Connect Ailoot to your favorite tools — AI API connections and client setup
Agents, editors, chats, and SDKs that accept an Ailoot key. The “Set up” button shows where to enter the URL and key, with a ready-to-copy configuration.
Category · API format
All integrations
Statistics
85applications
77ready
8under review
Terminal agents
Coding agents and CLIs: write code, run commands, and edit files.
Environment variables or the "env" block in ~/.claude/settings.json (for a single project, .claude/settings.local.json); VS Code extension — claudeCode.environmentVariables; check with the /status command
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/messages, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
export AILOOT_API_KEY="sk-..."
export ANTHROPIC_AUTH_TOKEN="$AILOOT_API_KEY"
claude
Example model: claude-sonnet-5-5
Important
Use a URL without /v1: Claude Code adds /v1/messages itself. ANTHROPIC_AUTH_TOKEN is sent as Authorization: Bearer, ANTHROPIC_API_KEY as x-api-key; Ailoot accepts both. CLAUDE_CODE_ENABLE_GATEWAY_MODEL_DISCOVERY=1 loads our models into /model through /v1/models. For 400 errors about unknown fields, use CLAUDE_CODE_DISABLE_EXPERIMENTAL_BETAS=1. Anthropic's documentation does not support using Claude Code with non-Claude models through a gateway, although our /v1/messages endpoint serves them; we recommend Claude models. Remote Control and voice input are unavailable when using a gateway.
~/.codex/config.toml (the user-level file only: model_provider/model_providers are ignored in a project’s .codex/config.toml); the key goes in the variable named by env_key
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/responses, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
~/.codex/config.toml
model = "gpt-6.1-sol"
model_provider = "ailoot"
[model_providers.ailoot]
name = "Ailoot"
base_url = "https://api.ailoot.ai/v1"
env_key = "AILOOT_API_KEY"
wire_api = "responses"
export AILOOT_API_KEY="sk-..."
codex
Example model: gpt-6.1-sol
Important
Codex uses only the OpenAI Responses API: the documentation lists only wire_api = "responses", so it needs our /v1/responses. Provider IDs openai, ollama, and lmstudio are reserved. Alternatively, use openai_base_url = "https://api.ailoot.ai/v1" with the built-in openai provider. Ailoot returns the model list in Codex manifest format through GET /v1/models with client_version. Choose a specific model whose catalog entry lists OpenAI Responses. In the 2026-09-30 snapshot, gemini-3.6-flash has OpenAI Chat, Anthropic Messages, and Gemini, but no Responses, so it is unsuitable for Codex. Check formats on the model card, not from the family name.
hermes model → “Custom endpoint (self-hosted / VLLM / etc.)” or ~/.hermes/config.yaml (the model section or providers dictionary)
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, POST /v1/responses, POST /v1/messages, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
The key is read from the variable AILOOT_API_KEY — set it before starting:
export AILOOT_API_KEY="sk-..." # add to ~/.zshrc or ~/.bashrc
setx AILOOT_API_KEY "sk-..." # then restart the terminal
Example model: claude-sonnet-5-5
Important
transport: chat_completions (OpenAI Chat), anthropic_messages (Anthropic Messages), or codex_responses (OpenAI Responses). LLM_MODEL in .env is no longer used; OPENAI_BASE_URL applies only to the openai-api provider. For anthropic_messages, use a URL without /v1; the Anthropic SDK appends /v1/messages. Auxiliary tasks (image recognition, history compaction) use the same main model by default.
OpenClaw configuration (~/.openclaw/openclaw.json) → models.providers.<id>; the default model is agents.defaults.model.primary; a safe update is openclaw config set models.providers.<id> '<json>' --strict-json --merge
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, POST /v1/responses, POST /v1/messages, POST /v1beta/models/{model}:generateContent, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
The key is read from the variable AILOOT_API_KEY — set it before starting:
export AILOOT_API_KEY="sk-..." # add to ~/.zshrc or ~/.bashrc
setx AILOOT_API_KEY "sk-..." # then restart the terminal
Example model: claude-sonnet-5-5
Important
api adapters include openai-completions (the default when baseUrl is set), openai-responses (only if the server supports /v1/responses, which Ailoot does), anthropic-messages, and google-generative-ai. OpenClaw’s anthropic-messages examples use baseUrl without /v1. OpenClaw can send images when the model’s input includes "image". The Claude examples keep text only: the 2026-09-30 catalog snapshot does not declare image input for these methods. Add "image" only for a specific model whose current catalog card declares image input in the selected API format. For example, gemini-3.6-flash declares image input for OpenAI Chat in that snapshot; actual support depends on the model card and format. The main configuration path follows common OpenClaw practice; the page reviewed also mentions ~/.openclaw/agents/<agentId>/agent/models.json.
opencode.json in the project or ~/.config/opencode/opencode.json → provider.<id>; set the key through /connect or "apiKey": "{env:VAR}"
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
The key is read from the variable AILOOT_API_KEY — set it before starting:
export AILOOT_API_KEY="sk-..." # add to ~/.zshrc or ~/.bashrc
setx AILOOT_API_KEY "sk-..." # then restart the terminal
Example model: claude-sonnet-5-5
Important
The @ai-sdk/openai-compatible package calls /v1/chat/completions. For models that require /v1/responses, opencode documentation specifies @ai-sdk/openai. A custom Anthropic-compatible provider with baseURL is not described in opencode’s documentation. Choose models whose current catalog entry lists OpenAI Chat. In the 2026-09-30 snapshot, gemini-3.6-flash has that format; check other Gemini models on their cards. This path does not use native Gemini format. A catalog method listing does not establish a live test of a specific client version.
OPENAI_API_KEY / OPENAI_BASE_URL / OPENAI_MODEL variables, /auth inside the session, or ~/.qwen/settings.json (modelProviders + security.auth.selectedType)
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, POST /v1/messages, POST /v1beta/models/{model}:generateContent, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
The key is read from the variable AILOOT_API_KEY — set it before starting:
export AILOOT_API_KEY="sk-..." # add to ~/.zshrc or ~/.bashrc
setx AILOOT_API_KEY "sk-..." # then restart the terminal
Example model: qwen3.8-max
Important
Protocol types: openai (Chat Completions), anthropic, gemini, and vertex-ai. For openai, include /v1 in baseUrl. The documentation examples for anthropic and gemini do not include baseUrl, so we recommend the openai option. For the openai type, choose a model whose catalog entry lists OpenAI Chat; gemini-3.6-flash has that format. Other Gemini models may support different formats, so check the specific model card. The settings.json option reads the key from AILOOT_API_KEY through envKey. Set it with export AILOOT_API_KEY=sk-... or an AILOOT_API_KEY=sk-... line in ~/.qwen/.env; otherwise, Qwen Code starts with an empty key and receives 401.
GEMINI_API_KEY and GOOGLE_GEMINI_BASE_URL environment variables; authentication method: ~/.gemini/settings.json → security.auth.selectedType = "gemini-api-key"; model: GEMINI_MODEL or the -m flag
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1beta/models/{model}:generateContent, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
Use an address without /v1beta: the SDK appends /v1beta/models/{model}:generateContent automatically. The address must use HTTPS. The key is sent in the x-goog-api-key header, which Ailoot accepts. Only models listed with the Gemini format in the catalog are suitable. Before running a request, check that the exact ID and this method are available to your key. Ailoot delivers paid Gemini responses in full after calculating the cost, not word by word, so text takes longer to appear for long responses.
~/.kimi/config.toml (providers and models sections, default_model); /login supports only Moonshot platforms, so set a custom URL manually in the file
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, POST /v1/responses, POST /v1/messages, POST /v1beta/models/{model}:generateContent, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
Types: openai_legacy (Chat Completions, base_url with /v1), openai_responses (Responses, with /v1), anthropic (without /v1), gemini (base_url is Google's root URL; for us, https://api.ailoot.ai). SearchWeb works only with Moonshot's own platform; with a different API URL, FetchURL switches to local downloading.
crushrc (Bash configuration): ./.crushrc, ./crushrc, or ~/.config/crush/crushrc; the legacy crush.json format is deprecated but still works
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, POST /v1/messages, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
The key is read from the variable AILOOT_API_KEY — set it before starting:
export AILOOT_API_KEY="sk-..." # add to ~/.zshrc or ~/.bashrc
setx AILOOT_API_KEY "sk-..." # then restart the terminal
Example model: deepseek-v4-pro
Important
The openai-compat type uses the OpenAI Chat format and a URL with /v1. The anthropic type uses the Messages format and a URL without /v1: Crush passes it to anthropic-sdk-go, which appends /v1/messages. The example in the Crush README with /v1 would produce /v1/v1/messages and a 404 error. The key is sent in the x-api-key header, which Ailoot accepts. If the model list is empty, the openai-compat provider fetches it from /v1/models. crushrc runs as code: do not launch Crush in someone else's directory containing this file. In this example, Claude has supports-images = false: the 2026-09-30 catalog snapshot lists only text for the selected Messages format. If your version of Crush supports this flag, enable it only for a model whose current catalog card lists image input in the selected API format; the client flag alone does not add image support to a model.
Environment variables OPENAI_API_BASE and OPENAI_API_KEY; the --model openai/<model> flag
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
The openai/ prefix is required: it tells Aider (through LiteLLM) to send requests to the OpenAI-compatible /v1/chat/completions endpoint. For models Aider does not recognize, it will warn that the context window is unknown — this is not an error. This documentation page does not cover an Anthropic endpoint with a custom base URL.
goose configure → Custom Providers → Add A Custom Provider, or JSON in ~/.config/goose/custom_providers/ (Windows: %APPDATA%\Block\goose\config\custom_providers\); alternatively, use the built-in OpenAI provider with OPENAI_HOST/OPENAI_BASE_PATH
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, POST /v1/messages, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
export AILOOT_API_KEY="sk-..."
goose session start --provider ailoot
# alternative using the built-in OpenAI provider:
export GOOSE_PROVIDER=openai
export OPENAI_HOST=https://api.ailoot.ai
export OPENAI_BASE_PATH=v1/chat/completions
export OPENAI_API_KEY="$AILOOT_API_KEY"
Example model: claude-sonnet-5-5
Important
Custom provider engines: openai, anthropic, ollama. In the Goose example, base_url for engine openai includes the full path to /v1/chat/completions. For the built-in provider, OPENAI_HOST is the root URL without a path, and OPENAI_BASE_PATH is the path (v1/chat/completions by default). Goose ignores keys in config.yaml; use an environment variable or the keyring. The repository has moved to aaif-goose/goose.
~/.factory/settings.json → customModels array; select a model with the /model command under “Custom models”
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, POST /v1/responses, POST /v1/messages, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
The key is read from the variable AILOOT_API_KEY — set it before starting:
export AILOOT_API_KEY="sk-..." # add to ~/.zshrc or ~/.bashrc
setx AILOOT_API_KEY "sk-..." # then restart the terminal
Example model: claude-sonnet-5-5
Important
provider: anthropic = Messages API (baseUrl as for Anthropic, without /v1), openai = Responses API (with /v1), generic-chat-completion-api = Chat Completions (with /v1). For anthropic, the key is sent in x-api-key by default (Ailoot accepts it); "authMode": "bearer" switches to Authorization. Keys remain local. Factory fully tests only the official Anthropic and OpenAI APIs.
Settings → Model Routing → Add → Custom URL (or workspace Model Routing for a team); from the terminal — amp config model-providers (list, test, set-priority, edit-router…)
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, POST /v1/responses, POST /v1/messages, POST /v1beta/models/{model}:generateContent, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
Model Routing → Add → Custom URL
Format: anthropic-messages
Base URL: https://api.ailoot.ai
API key: sk-...
Model mapping:
anthropic/claude-fable-5 -> anthropic/claude-fable-5
Second connection for GPT:
Format: responses
Base URL: https://api.ailoot.ai/v1
API key: sk-...
Model mapping:
openai/gpt-5.6-sol -> gpt-5.6-sol
Check: amp config model-providers check-access --provider-model anthropic/claude-fable-5
Example model: claude-fable-5
Important
Custom URL formats: chat-completions (<base>/chat/completions), responses (<base>/responses), anthropic-messages (<base>/v1/messages; /v1 is not duplicated), google-genai (<base>/models/<model>:streamGenerateContent). The key is sent as Authorization: Bearer. Mappings use canonical Amp IDs in the provider/model form; the arrow specifies the ID sent to us. Amp handles models not covered by the connection itself, using its own credits. The documentation requires the custom host to be accessible from Amp — our URL is public, so this requirement is met.
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, POST /v1/messages, POST /v1beta/models/{model}:generateContent, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
In AIChat's example configuration, api_base for claude includes /v1, while gemini uses /v1beta; use the same paths for Ailoot. The three clients provide access to all three Ailoot formats: OpenAI, Anthropic, and Gemini.
Continue YAML configuration (config.yaml), passed to the CLI with --config <path>
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
For OpenAI-compatible services, Continue uses provider: openai and an apiBase with /v1. The configuration format is shared by the extension and CLI; pass the file with the --config flag.
forge provider login → OpenAI-compatible (recommended); the deprecated option uses .env with OPENAI_URL and OPENAI_API_KEY, plus model in forge.yaml
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
Configuration through .env is marked as deprecated: on first launch, Forge migrates the variables to its file storage. The README does not list the exact labels in the provider login wizard.
COPILOT_PROVIDER_BASE_URL, COPILOT_PROVIDER_TYPE, COPILOT_PROVIDER_API_KEY, and COPILOT_MODEL environment variables (or the --model flag)
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, POST /v1/messages, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
Types: openai (default, any OpenAI-compatible server), azure, anthropic. The model must support tool calling and streaming output. COPILOT_PROVIDER_WIRE_API selects the protocol, but the page does not list its values. Limits can be set with COPILOT_PROVIDER_MAX_PROMPT_TOKENS / MAX_OUTPUT_TOKENS.
~/.config/gptme/config.toml → [[providers]]; select a model as <provider>/<model>
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
~/.config/kilo/kilo.jsonc (or ./kilo.jsonc; KILO_CONFIG is also supported) → provider.<id>.options.baseURL; in the extension: Settings → Providers → Custom provider
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, POST /v1/responses, POST /v1/messages, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
The key is read from the variable AILOOT_API_KEY — set it before starting:
export AILOOT_API_KEY="sk-..." # add to ~/.zshrc or ~/.bashrc
setx AILOOT_API_KEY "sk-..." # then restart the terminal
Example model: claude-sonnet-5-5
Important
The Custom provider dialog lets you choose a protocol: OpenAI Compatible (Chat Completions), OpenAI Responses, or Anthropic Messages. {env:VAR} works only in the trusted global configuration (~/.config/kilo); the project-level kilo.json will not substitute a key from an environment variable.
The extra-openai-models.yaml file in llm's configuration directory; set the key with llm keys set <name>
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, POST /v1/responses, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
Chat Completions is the default; responses: true switches the model to /responses. When api_base is set, the OpenAI key is not sent, so api_key_name is required. This is a CLI utility for making requests, not an agent.
Agent YAML configuration → model section (model_name, model_kwargs); MSWEA_MINI_CONFIG_PATH changes the default configuration path
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
Uses LiteLLM: model_kwargs are passed directly to litellm.completion. Without a price registry, LiteLLM reports an unknown cost; disable tracking with cost_tracking: "ignore_errors" or MSWEA_COST_TRACKING=ignore_errors. OPENAI_API_KEY is LiteLLM’s standard key variable, but the mini-SWE-agent page does not show it.
Ctrl+P → Settings (provider and model), ~/.openhands/agent_settings.json (llm.model / llm.api_key / llm.base_url), or LLM_MODEL, LLM_API_KEY, and LLM_BASE_URL variables with --override-with-envs
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
export AILOOT_API_KEY="sk-..."
# for a single run using environment variables:
export LLM_MODEL=openai/claude-sonnet-5-5
export LLM_API_KEY="$AILOOT_API_KEY"
export LLM_BASE_URL=https://api.ailoot.ai/v1
openhands --override-with-envs
Example model: openai/claude-sonnet-5-5
Important
OpenHands uses LiteLLM. For an OpenAI proxy, the documentation specifies a Custom Model such as openai/<model> and the proxy’s Base URL. Changes made through --override-with-envs are not saved. In the web interface: Settings → LLM → Advanced → Custom Model, Base URL, API Key.
~/.pi/agent/models.json, or the project’s .pi/models.json after trusting the project; select a model with /model
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, POST /v1/messages, POST /v1beta/models/{model}:generateContent, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
The key is read from the variable AILOOT_API_KEY — set it before starting:
export AILOOT_API_KEY="sk-..." # add to ~/.zshrc or ~/.bashrc
setx AILOOT_API_KEY "sk-..." # then restart the terminal
Example model: claude-sonnet-5-5
Important
Pi documentation explicitly supports OpenAI-, Anthropic-, and Google-compatible endpoints through models.json; the example uses api: openai-completions. apiKey accepts $NAME/${NAME}, a literal value, or !command. The file is reread when /model opens.
plandex models custom, or \models custom in the REPL, opens the JSON file for custom models and providers
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
The key is read from the variable AILOOT_API_KEY — set it before starting:
export AILOOT_API_KEY="sk-..." # add to ~/.zshrc or ~/.bashrc
setx AILOOT_API_KEY "sk-..." # then restart the terminal
Example model: claude-sonnet-5-5
Important
Custom providers work only in self-hosted Plandex. In the cloud, BYO keys are limited to built-in providers. All custom providers must be OpenAI-compatible (Chat Completions). Then add the model to a custom model pack.
trae_config.yaml → model_providers.<name>.base_url; alternatively, OPENAI_BASE_URL / ANTHROPIC_BASE_URL variables in .env
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, POST /v1/messages, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
Use spaces only in YAML, not tabs. For Anthropic format, use provider: anthropic and ANTHROPIC_BASE_URL. The README does not specify whether anthropic expects /v1; use https://api.ailoot.ai.
Settings → search for “inference endpoint” → Custom inference endpoint: base URL, key, model IDs; then select the model in the picker
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
Settings → inference endpoint
Endpoint URL: https://api.ailoot.ai/v1
API key: sk-...
Models: claude-sonnet-5-5, gpt-6.1-sol, deepseek-v4-pro
Then select a model from the list, for example claude-sonnet-5-5 (not Auto).
Example model: claude-sonnet-5-5
Important
OpenAI Chat Completions only (POST /v1/chat/completions). Requests pass through Warp’s servers: the key is stored locally but sent to the Warp backend with every request. Auto mode and custom routers always use Warp credits; cloud agents do not use this endpoint. Available on Free and paid plans for individuals and teams of up to 10 people. Choose models whose current catalog entry lists OpenAI Chat. In the 2026-09-30 snapshot, gemini-3.6-flash has this format; check other Gemini models on their cards. This path does not use native Gemini format. A catalog method listing does not establish a live test of a specific client version.
Cline panel → gear icon (Settings) → API Provider = “OpenAI Compatible” → Base URL, API Key, Model ID fields. Another option: API Provider = “Anthropic” → check “Use custom base URL”.
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, POST /v1/messages, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
OpenAI Compatible (recommended):
API Provider: OpenAI Compatible
Base URL: https://api.ailoot.ai/v1
API Key: sk-...
Model ID: claude-sonnet-5-5 (or gpt-6.1-sol, deepseek-v4-pro, qwen3.8-max)
Model Configuration: Context Window and Max Output Tokens — match the model; Image Support — only for models that can process images.
Anthropic (Messages):
API Provider: Anthropic
Anthropic API Key: sk-...
[x] Use custom base URL → https://api.ailoot.ai
Example model: claude-sonnet-5-5
Important
OpenAI Compatible calls /v1/chat/completions, so the URL includes /v1. The Anthropic option calls /v1/messages and uses a URL without /v1, but its model list contains only Anthropic names; other Ailoot models are available through OpenAI Compatible. Agent mode requires a model that supports tool calling. Gemini 3.1 Pro models are not available through chat/completions (they support only the Gemini and Anthropic formats); use gemini-3.6-flash here.
Roo Code panel → gear icon → API Provider = “OpenAI Compatible” → Base URL, API Key, Model; optionally Model Configuration (Max Output Tokens, Context Window, Image Support, Computer Use, prices). The Anthropic provider has a “Use custom base URL” checkbox.
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, POST /v1/messages, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
API Provider: OpenAI Compatible
Base URL: https://api.ailoot.ai/v1
API Key: sk-...
Model: gpt-6.1-sol (or claude-sonnet-5-5, kimi-k3, glm-5.3)
Alternative via Anthropic:
API Provider: Anthropic, API key sk-..., [x] Use custom base URL → https://api.ailoot.ai
Example model: gpt-6.1-sol
Important
Roo Code calls tools only in OpenAI format, so the model must support tools in /v1/chat/completions. The Anthropic option calls /v1/messages and uses a URL without /v1.
Settings (gear icon) → Providers tab → “Custom provider” at the bottom → Provider ID, Display name, Provider API = “OpenAI Compatible”, Base URL, API key, Models (the list is fetched from /v1/models). Settings are saved in kilo.json.
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
Provider ID: ailoot
Display name: Ailoot
Provider API: OpenAI Compatible
Base URL: https://api.ailoot.ai/v1
API key: sk-...
Models: select from the list, for example claude-sonnet-5-5, gpt-6.1-sol, deepseek-v4-pro
Example model: deepseek-v4-pro
Important
Kilo requests {Base URL}/models and displays the model list automatically; our GET /v1/models returns the full catalog. You can also specify the full .../v1/chat/completions URL. Custom headers are supported.
The ~/.continue/config.yaml file (or Continue's configuration editor): an entry in models[] with provider: openai and apiBase.
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
name: Ailoot
version: 0.0.1
schema: v1
models:
- name: Ailoot Claude Sonnet 5.5
provider: openai
model: claude-sonnet-5-5
apiBase: https://api.ailoot.ai/v1
apiKey: sk-...
- name: Ailoot GPT-6.1 Sol
provider: openai
model: gpt-6.1-sol
apiBase: https://api.ailoot.ai/v1
apiKey: sk-...
# useResponsesApi: false # only if you need to force requests to /chat/completions
Example model: claude-sonnet-5-5
Important
apiBase must end with /v1. For gpt-5/o-series names, Continue may automatically choose /v1/responses; our gpt-* models support both /v1/responses and /v1/chat/completions, while useResponsesApi: false forces chat/completions. useLegacyCompletionsEndpoint is not needed.
Agent Settings (agent::OpenSettings) → LLM Providers → Add Provider, or settings.json → language_models.openai_compatible / language_models.anthropic_compatible. Set the key in the settings window, where it is stored in the system keychain, or through <PROVIDER_ID_UPPERCASE>_API_KEY.
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, POST /v1/responses, POST /v1/messages, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
openai_compatible uses a URL with /v1 and calls /v1/chat/completions; capabilities.chat_completions = false switches the model to /v1/responses. anthropic_compatible uses a URL without /v1 and calls /v1/messages; Zed sends x-api-key, which Ailoot accepts. images = false and parallel_tool_calls = false by default. Zed can send images, but enable images only for a specific model whose current catalog card declares image input in the selected API format. The GPT and Claude examples use images = false: the 2026-09-30 snapshot lists text only for them. For example, gemini-3.6-flash declares image input for OpenAI Chat in that snapshot; actual support depends on the model card and format. Reasoning models have a reasoning_effort field. The example’s max_tokens values represent context size; check exact limits in the model catalog.
Chat model picker → Manage Language Models (gear icon) → Add Models → Custom Endpoint → VS Code opens chatLanguageModels.json. The old github.copilot.chat.customOAIModels setting is deprecated.
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, POST /v1/responses, POST /v1/messages, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
url is the FULL endpoint path. apiType values: chat-completions, responses (url .../v1/responses), messages (url .../v1/messages). BYOK works without a Copilot plan or GitHub sign-in, but only for chat/agent; inline suggestions stay on Copilot. Organization admins (Business/Enterprise) can disable the BYOK policy. The token limits above are examples; use catalog values. These examples use vision = false: the 2026-09-30 catalog snapshot lists text only for GPT/OpenAI Chat and Claude/Anthropic Messages. If your VS Code version offers vision, enable it only for a model whose current catalog card declares image input in the selected API format. A client field does not establish image support for the selected model.
Settings | Tools | AI Assistant | Providers & API keys → Third-party AI providers → OpenAI-compatible → URL, API Key → Test Connection → Apply → AI Chat / Models Assignment
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
Provider: OpenAI-compatible
URL: https://api.ailoot.ai/v1
API Key: sk-...
Example model: claude-sonnet-5-5
Important
Choose OpenAI-compatible and set URL to https://api.ailoot.ai/v1 with your API Key. Click Test Connection, then Apply, and select a model in AI Chat. Assign it in Models Assignment for IDE features. Enable Tool calling only when the chosen model supports tools; AI completion has separate settings.
Xcode → Settings → Intelligence → Chat → Add a Chat Provider → Internet Hosted → URL, API Key, API Key Header, description → Add.
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
Type: Internet Hosted
URL: https://api.ailoot.ai (WITHOUT /v1: Xcode adds /v1/models and /v1/chat/completions itself)
API Key: sk-...
API Key Header: x-api-key
Description: Ailoot
→ then select models from the list (for example claude-sonnet-5-5, gpt-6.1-sol)
Example model: claude-sonnet-5-5
Important
Apple requires the provider to support Chat Completions and respond at {URL}/v1/models and {URL}/v1/chat/completions. Ailoot accepts x-api-key and Authorization: Bearer. Reports suggest Xcode adds the Bearer prefix itself when Authorization is selected. Xcode agents are configured separately and are outside this guide.
opts for yetone/avante.nvim in lazy.nvim: opts.providers.<name> with __inherited_from = 'openai' (a custom OpenAI-compatible provider), or override providers.claude.endpoint; the key comes from the variable specified in api_key_name (AVANTE_* names are supported).
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, POST /v1/messages, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
opts = {
provider = "ailoot",
providers = {
ailoot = {
__inherited_from = "openai",
endpoint = "https://api.ailoot.ai/v1",
api_key_name = "AILOOT_API_KEY",
model = "claude-sonnet-5-5",
},
claude = {
endpoint = "https://api.ailoot.ai", -- Anthropic Messages, without /v1
model = "claude-sonnet-5-5",
},
},
}
-- in the terminal: export AILOOT_API_KEY=sk-...
-- the claude provider reads AVANTE_ANTHROPIC_API_KEY or ANTHROPIC_API_KEY
Example model: claude-sonnet-5-5
Important
In the plugin's wiki, the OpenAI-compatible URL ends with /v1 (the plugin appends /chat/completions). The built-in claude provider uses a URL such as https://api.anthropic.com, so use a URL without /v1 for Ailoot. Custom providers without __inherited_from require a full URL and their own response parsing functions.
require('codecompanion').setup({ adapters = { http = { ... } } }) using adapters.extend('openai_compatible', ...) or extend('anthropic', ...); then interactions.chat/inline.adapter.
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, POST /v1/messages, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
The key is read from the variable AILOOT_API_KEY — set it before starting:
export AILOOT_API_KEY="sk-..." # add to ~/.zshrc or ~/.bashrc
setx AILOOT_API_KEY "sk-..." # then restart the terminal
Example model: gpt-6.1-sol
Important
openai_compatible separates the host (env.url, without /v1) from the path (env.chat_url). Values in env can be an environment variable name, cmd:..., file:..., or a Lua function. The anthropic adapter can similarly be extended with a custom url for /v1/messages.
init.el: register a backend with gptel-make-openai (OpenAI Chat), gptel-make-anthropic (Messages), or gptel-make-gemini; each has :host, :endpoint, :key, :stream, :models. Then set gptel-backend and gptel-model.
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, POST /v1/messages, POST /v1beta/models/{model}:generateContent, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
The key is read from the variable AILOOT_API_KEY — set it before starting:
export AILOOT_API_KEY="sk-..." # add to ~/.zshrc or ~/.bashrc
setx AILOOT_API_KEY "sk-..." # then restart the terminal
Example model: claude-sonnet-5-5
Important
:host is only the hostname (without https:// or a path); set the path in :endpoint. gptel-make-anthropic sends the x-api-key and anthropic-version headers, which Ailoot accepts.
Preferences → AI-features → OpenAiCustom, or the ai-features.openAiCustom.customOpenAiModels key in settings.json; open AI Configuration with Alt+A.
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
Settings → Models → Add Model → Custom Model → API Format, Request URL (Full URL toggle), Model ID, Display Name, API Key; optionally Advanced Settings.
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, POST /v1/messages, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
OpenAI format:
API Format: OpenAI Chat Completions
Full URL: OFF
Request URL: https://api.ailoot.ai/v1
Model ID: gpt-6.1-sol
API Key: sk-...
Anthropic format:
API Format: Anthropic Messages
Full URL: OFF
Request URL: https://api.ailoot.ai
Model ID: claude-sonnet-5-5
API Key: sk-...
Example model: gpt-6.1-sol
Important
With Full URL disabled, Trae appends the selected format’s path: include /v1 for OpenAI and omit it for Anthropic. With Full URL enabled, enter https://api.ailoot.ai/v1/chat/completions or https://api.ailoot.ai/v1/messages. Model Series adapts hints to the model family.
Settings > Admin > Connections > Manage OpenAI API Connections > + Add Connection, or OPENAI_API_BASE_URL / OPENAI_API_KEY environment variables when starting the container
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
UI: URL = https://api.ailoot.ai/v1
API Key = sk-...
Model IDs: leave blank (loaded from /v1/models) or manually add claude-sonnet-5-5, gpt-6.1-sol
Docker env:
OPENAI_API_BASE_URL=https://api.ailoot.ai/v1
OPENAI_API_KEY=sk-...
Example model: claude-sonnet-5-5
Important
Uses OpenAI Chat Completions. The URL must include /v1 and have no trailing slash. The model list comes from /v1/models. For a large catalog (252 models), limiting Model IDs and enabling model-list caching is more convenient: Cache Base Model List / ENABLE_BASE_MODELS_CACHE=True. RAG embeddings use a separate RAG_OPENAI_API_BASE_URL.
librechat.yaml file → endpoints.custom (key in .env via ${VAR}); for the Anthropic format, use provider: 'anthropic'
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, POST /v1/messages, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
A standard custom endpoint uses OpenAI Chat Completions (baseURL with /v1). provider: 'anthropic' enables the native /v1/messages client. The documentation's Anthropic example specifies baseURL without /v1 (https://api.anthropic.com), so use https://api.ailoot.ai for Ailoot. fetch: true retrieves the list from /v1/models.
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, POST /v1/responses, POST /v1/messages, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
Provider Name: Ailoot
Type: OpenAI (or OpenAI Responses / Anthropic)
API Key: sk-...
Base URL: https://api.ailoot.ai
→ check the preview of the final URL below the field: it should be https://api.ailoot.ai/v1/chat/completions (for Anthropic: https://api.ailoot.ai/v1/messages)
Models: + claude-sonnet-5-5, gpt-6.1-sol
Then click Model Check
Example model: claude-sonnet-5-5
Important
The documentation confirms support for the OpenAI, Anthropic, OpenAI Responses, and Gemini protocols, and a preview of the final path below the field, but it does not describe the rules for appending /v1 — use the preview as your guide. All three of our formats (Chat Completions, Responses, Messages) are compatible.
Settings (sidebar button) > Model Provider > Add (at the bottom) > type: OpenAI API Compatible
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
Name: Ailoot
API Mode: OpenAI API Compatible
API Key: sk-...
API Host: https://api.ailoot.ai
API Path: (blank, defaults to /v1/chat/completions)
Models: + gpt-6.1-sol, claude-sonnet-5-5 (vision is disabled in these examples; tool_use depends on the capabilities of the selected model)
Example model: gpt-6.1-sol
Important
The documentation lists API Host without /v1 (example: https://api.openai.com), with /v1/chat/completions as the default path. There is also a Google Gemini compatible type, but it has not been tested with our /v1beta. JSON configuration import (type: 'openai', apiHost, apiPath) is available from 1.15.1 via chatbox://provider/import?config=<base64>. Vision is disabled for the GPT and Claude examples shown: the 2026-09-30 catalog snapshot lists only text for the OpenAI Chat format. If your Chatbox version offers vision, enable it only for a model whose current catalog card lists image input in the selected API format; a client's ability to attach an image does not prove that the model supports it.
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, POST /v1/messages, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
OPENAI_API_KEY=sk-...
OPENAI_PROXY_URL=https://api.ailoot.ai/v1
OPENAI_MODEL_LIST=-all,+gpt-6.1-sol,+claude-sonnet-5-5,+deepseek-v4-pro
# Anthropic format (optional)
ANTHROPIC_API_KEY=sk-...
ANTHROPIC_PROXY_URL=https://api.ailoot.ai
Example model: claude-sonnet-5-5
Important
OPENAI_PROXY_URL defaults to https://api.openai.com/v1, so include /v1. ANTHROPIC_PROXY_URL defaults to https://api.anthropic.com, without /v1. GOOGLE_PROXY_URL can in principle connect Gemini format through https://api.ailoot.ai, but we have not tested this option. Use server mode: requests sent directly from the browser may currently be rejected by Ailoot’s CORS rules.
Settings > Model Providers > Add Provider → choose the OpenAI-compatible or Anthropic-compatible format
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, POST /v1/messages, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
OpenAI-compatible:
Provider name: Ailoot
Base URL: https://api.ailoot.ai/v1
API key: sk-...
Anthropic-compatible:
Provider name: Ailoot Claude
Base URL: https://api.ailoot.ai
API key: sk-...
Models: loaded from {base_url}/models or + manually: claude-sonnet-5-5
Example model: claude-sonnet-5-5
Important
Docs: for OpenAI-compatible providers, Base URL must include /v1 (otherwise you get 404); a trailing slash is removed. The Anthropic-mode example for LiteLLM omits /v1. Model capabilities (tools, vision) must be configured manually for custom providers. Backup keys are supported (retries only on 401/403/429).
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
Custom Endpoint (Base URL): https://api.ailoot.ai/v1
Custom API Key: sk-...
Model: select from the list (/v1/models) or enter claude-sonnet-5-5
Example model: claude-sonnet-5-5
Important
The documentation says to add /v1 when troubleshooting a connection and NOT to add /chat/completions. Models are fetched if the server provides /v1/models, which Ailoot does.
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
UI: Base URL = https://api.ailoot.ai/v1
API Key = sk-...
Chat Model Name = deepseek-v4-pro
Token context window = according to the model
Docker .env:
LLM_PROVIDER='generic-openai'
GENERIC_OPEN_AI_BASE_PATH='https://api.ailoot.ai/v1'
GENERIC_OPEN_AI_MODEL_PREF='deepseek-v4-pro'
GENERIC_OPEN_AI_MODEL_TOKEN_LIMIT=128000
GENERIC_OPEN_AI_API_KEY='sk-...'
Example model: deepseek-v4-pro
Important
The documentation labels this provider “for developers”. If needed, use GENERIC_OPEN_AI_MODEL_MAX_TOKEN_KEY='max_completion_tokens'. Embeddings have a separate generic-openai embedder (EMBEDDING_BASE_PATH + GENERIC_OPEN_AI_EMBEDDING_API_KEY) — the nemotron-3-embed model through /v1/embeddings.
Environment variables at deployment: OPENAI_API_KEY / OPENAI_API_HOST, ANTHROPIC_API_KEY / ANTHROPIC_API_HOST; in the UI — Models > Add OpenAI-compatible
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, POST /v1/messages, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
The documentation describes these variables as overrides for the OpenAI and Anthropic backend host (for proxies and custom endpoints). It says “host”, so we use a URL without /v1, but the documentation does not specify the exact path format — check it on the first request.
Settings > Models > (+) > OpenAI-compatible Server
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
Friendly name: Ailoot
API Endpoint URL: https://api.ailoot.ai/v1/chat/completions
API Key: sk-...
Model ID: gpt-6.1-sol
Context length: according to the model
Streaming: enable
Example model: gpt-6.1-sol
Important
A full chat completions URL is required. To use it with AI Command, make this service the default. These instructions follow the BoltAI help pages; screens may differ in BoltAI 2.
The .env.local file (OPENAI_BASE_URL, OPENAI_API_KEY)
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
Model Hub > Model Providers > Add Provider > OpenAI-compatible
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
Provider: OpenAI Compatible
Name: Ailoot
API Endpoint: https://api.ailoot.ai/v1
API Key: sk-...
Models: gpt-6.1-sol, claude-sonnet-5-5
Example model: gpt-6.1-sol
Important
The documentation confirms a custom OpenAI-compatible provider but does not specify the URL format. /v1 is the standard OpenAI API root; if you receive 404, check the URL without /v1. A custom Anthropic-compatible endpoint is not described.
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
Use OpenAI Chat through the NextChat server for this setup. When deploying on Vercel or Docker, set BASE_URL=https://api.ailoot.ai without /v1, OPENAI_API_KEY, CUSTOM_MODELS and DEFAULT_MODEL. Redeploy the server after changing these variables.
Settings > Copilot > BYOK > Add a provider > Add a custom provider
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
Base URL: https://api.ailoot.ai/v1
API key: sk-...
Model ID: claude-sonnet-5-5 (or select from the /v1/models list)
Then Test; enable the model in Basic > Agents > Quick Chat
Example model: claude-sonnet-5-5
Important
The model list comes from /v1/models, and you can enter a model manually. If Test passes but Quick Chat does not send messages, turn on Enable CORS for the provider; the response will arrive in full without streaming. Keys are stored in Obsidian Keychain.
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
Provider: Custom
Name: Ailoot
Base URL: https://api.ailoot.ai/v1
API Key: sk-...
Then manually add the models: gpt-6.1-sol, claude-sonnet-5-5
Example model: gpt-6.1-sol
Important
The documentation confirms Custom with an API URL and key but does not specify the URL format. /v1 is the standard OpenAI API root. Automatic model discovery is guaranteed only for Ollama, LM Studio, and Llamafile; other providers require models to be added manually.
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
base_url is an OpenAI-compatible endpoint without the /chat/completions suffix. Raycast rereads the file without restarting. If abilities is not set, tools and vision are treated as unsupported. Models appear in Quick AI, AI Chat, and AI Commands. This example uses vision.supported = false: the 2026-09-30 catalog snapshot lists text only for Claude/OpenAI Chat. If your Raycast version supports this setting, enable vision only for a model whose current catalog card declares image input in OpenAI Chat format; the client flag alone does not prove API image support.
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
Name: Claude Sonnet 5.5 (Ailoot)
API Type: OpenAI Chat Completions
Endpoint: https://api.ailoot.ai/v1/chat/completions
Model ID: claude-sonnet-5-5
Custom Headers: Authorization = Bearer sk-...
Context Length: according to the model
Test → Save
Example model: claude-sonnet-5-5
Important
Enter Endpoint as the full path. The key is sent through a custom Authorization: Bearer sk-... header, which Ailoot accepts. TypingMind’s web version calls the API directly from the browser, and CORS may currently block these requests; the desktop version is more reliable. TypingMind’s Gemini type sends the key in apiKey, which Ailoot does not accept. Use OpenAI Chat Completions instead.
OpenAI(base_url=..., api_key=...) constructor parameters, or OPENAI_BASE_URL and OPENAI_API_KEY environment variables
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, POST /v1/responses, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
export AILOOT_API_KEY="sk-..."
# environment variables
export OPENAI_BASE_URL="https://api.ailoot.ai/v1"
export OPENAI_API_KEY="$AILOOT_API_KEY"
# or in code
ailoot-openai-python-sdk.py
import os
from openai import OpenAI
client = OpenAI(max_retries=0, base_url="https://api.ailoot.ai/v1", api_key=os.environ["AILOOT_API_KEY"])
r = client.chat.completions.create(model="gpt-6.1-sol", messages=[{"role":"user","content":"Hello"}])
print(r.choices[0].message.content)
# Responses API: client.responses.create(model="gpt-6.1-sol", input="Hello")
# Embeddings: client.embeddings.create(model="nemotron-3-embed", input=["text"])
python3 ailoot-openai-python-sdk.py
Example model: gpt-6.1-sol, claude-sonnet-5-5, deepseek-v4-pro, qwen3.8-max (chat); nemotron-3-embed (embeddings)
Important
Include /v1 in the URL: the SDK appends /chat/completions, /responses, /embeddings, /images/*, and /audio/*. Chat Completions and Responses work; Responses requires a model whose catalog entry lists that format. Streaming and tools use the standard formats. Ailoot errors are JSON with a Request ID and are caught as openai.APIStatusError.
Constructor option new OpenAI({ baseURL, apiKey }); the SDK reads OPENAI_BASE_URL by default
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, POST /v1/responses, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
The key is read from the variable AILOOT_API_KEY — set it before starting:
export AILOOT_API_KEY="sk-..." # add to ~/.zshrc or ~/.bashrc
setx AILOOT_API_KEY "sk-..." # then restart the terminal
Example model: claude-sonnet-5-5, gpt-6.1-sol, grok-4.7
Important
The README shows baseURL. Reading OPENAI_BASE_URL is visible in the client source, src/client.ts; its default is https://api.openai.com/v1. Include /v1 in the Ailoot URL. Both chat.completions and responses are available.
The Anthropic(base_url=...) parameter or ANTHROPIC_BASE_URL environment variable; key: api_key or ANTHROPIC_API_KEY
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/messages, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
Example model: claude-sonnet-5-5, claude-opus-5-5; also via /v1/messages gpt-6.1-sol, deepseek-v4-pro, gemini-3.6-flash
Important
Use a URL WITHOUT /v1: the SDK adds /v1/messages itself. The gateway accepts the key in x-api-key (as sent by the SDK) and Authorization: Bearer. Streaming, tools, and /v1/messages/count_tokens work. Batch processing (Batches) and the Files API are not listed as supported in the Ailoot catalog.
Providers: createOpenAICompatible({ name, baseURL, apiKey }) from @ai-sdk/openai-compatible; createAnthropic({ baseURL, apiKey }) from @ai-sdk/anthropic; createGoogleGenerativeAI({ baseURL, apiKey }) from @ai-sdk/google; createOpenAI({ baseURL }) or OPENAI_BASE_URL
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, POST /v1/responses, POST /v1/messages, POST /v1beta/models/{model}:generateContent, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
The key is read from the variable AILOOT_API_KEY — set it before starting:
export AILOOT_API_KEY="sk-..." # add to ~/.zshrc or ~/.bashrc
setx AILOOT_API_KEY "sk-..." # then restart the terminal
Example model: deepseek-v4-pro (OpenAI-compatible), claude-sonnet-5-5 (Anthropic), gemini-3.6-flash (Gemini)
Important
Providers use different URLs: OpenAI-compatible and Anthropic use /v1 (Anthropic defaults to https://api.anthropic.com/v1); Google uses /v1beta (default https://generativelanguage.googleapis.com/v1beta). The @ai-sdk/openai provider defaults to the Responses API. For models without responses, use openai.chat("id") or createOpenAICompatible.
ChatOpenAI(base_url=...) or the OPENAI_API_BASE / OPENAI_BASE_URL environment variables; ChatAnthropic(base_url=...) or ANTHROPIC_API_URL / ANTHROPIC_BASE_URL
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, POST /v1/responses, POST /v1/messages, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
import os
from langchain_openai import ChatOpenAI
llm = ChatOpenAI(model="gpt-6.1-sol", base_url="https://api.ailoot.ai/v1", api_key=os.environ["AILOOT_API_KEY"])
print(llm.invoke("Hello").content)
from langchain_anthropic import ChatAnthropic
claude = ChatAnthropic(model="claude-sonnet-5-5", base_url="https://api.ailoot.ai", api_key=os.environ["AILOOT_API_KEY"])
Example model: gpt-6.1-sol, qwen3.8-max (ChatOpenAI); claude-sonnet-5-5 (ChatAnthropic)
Important
ChatOpenAI: use a URL with /v1. According to the documentation, ChatOpenAI supports only the official OpenAI specification: nonstandard fields such as reasoning_content from third-party models are not retained. For Claude, prefer ChatAnthropic: use a URL without /v1, as the SDK adds /v1/messages automatically. ChatAnthropic's base_url parameter and ANTHROPIC_API_URL/ANTHROPIC_BASE_URL environment variables are documented in the class documentation in the source code (langchain_anthropic/chat_models.py), but not on the integration page. The use_responses_api flag enables Responses mode in ChatOpenAI.
crewai.LLM(model=..., base_url=..., api_key=...); OPENAI_API_KEY and OPENAI_BASE_URL environment variables; for Claude, the native anthropic/ provider with base_url
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, POST /v1/messages, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
import os
from crewai import LLM
# OpenAI-compatible format
llm = LLM(model="openai/gpt-6.1-sol", base_url="https://api.ailoot.ai/v1", api_key=os.environ["AILOOT_API_KEY"])
# Anthropic Messages
claude = LLM(model="anthropic/claude-sonnet-5-5", base_url="https://api.ailoot.ai", api_key=os.environ["AILOOT_API_KEY"], max_tokens=4096)
Example model: openai/gpt-6.1-sol, anthropic/claude-sonnet-5-5
Important
The openai/ or anthropic/ prefix selects the provider in CrewAI; the request sends the id without the prefix. For OpenAI, the URL includes /v1; for Anthropic, it does not. The documentation also describes the custom_openai=True flag for an arbitrary OpenAI-compatible gateway. Anthropic requires max_tokens.
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1beta/models/{model}:generateContent, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
import os
from google import genai
from google.genai import types
client = genai.Client(
api_key=os.environ["AILOOT_API_KEY"],
http_options=types.HttpOptions(base_url="https://api.ailoot.ai"))
r = client.models.generate_content(model="gemini-3.6-flash", contents="Hello")
print(r.text)
Example model: gemini-3.6-flash, gemini-3.1-pro
Important
Use a URL without /v1beta: the SDK builds /v1beta/models/{model}:generateContent. The key is sent in the x-goog-api-key header, which Ailoot accepts. Only models supporting the Gemini format are available. The documentation for the JavaScript version (@google/genai) does not describe a custom endpoint, so it is not included here.
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
Example model: openai/claude-sonnet-5-5, openai/qwen3.8-max, openai/nemotron-3-embed
Important
The openai/ prefix selects the OpenAI-compatible protocol; Ailoot receives the id without the prefix. LiteLLM adds /chat/completions and /embeddings automatically. According to the documentation, if you get a Not Found error, the URL should end with /v1.
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
The key is read from the variable AILOOT_API_KEY — set it before starting:
export AILOOT_API_KEY="sk-..." # add to ~/.zshrc or ~/.bashrc
setx AILOOT_API_KEY "sk-..." # then restart the terminal
Example model: openai/claude-sonnet-5-5, openai/gpt-6.1-sol
Important
The proxy is useful if you want to issue your team internal keys backed by a single Ailoot key. Charges in Ailoot are attributed to the key specified in api_key. Use a URL with /v1.
llama-index-llms-openai-like package, OpenAILike(api_base=..., api_key=..., model=..., is_chat_model=True) class
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
pip install llama-index-llms-openai-like
import os
from llama_index.llms.openai_like import OpenAILike
llm = OpenAILike(
model="deepseek-v4-pro",
api_base="https://api.ailoot.ai/v1",
api_key=os.environ["AILOOT_API_KEY"],
is_chat_model=True,
is_function_calling_model=True,
context_window=128000)
print(llm.complete("Hello"))
Example model: deepseek-v4-pro, claude-sonnet-5-5, gpt-6.1-sol
Important
Use a URL with /v1. is_chat_model=True is required; otherwise, the class uses the legacy /completions endpoint. Set is_function_calling_model=True so agents can use tools. Set context_window according to the model card in the Ailoot catalog.
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, POST /v1/responses, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
import { Agent } from "@mastra/core/agent";
const agent = new Agent({
id: "ailoot-agent",
name: "Ailoot Agent",
instructions: "Respond in Russian",
model: {
id: "ailoot/claude-sonnet-5-5",
url: "https://api.ailoot.ai/v1",
apiKey: process.env.AILOOT_API_KEY
// api: "responses" // only for models with the responses method
}
});
The key is read from the variable AILOOT_API_KEY — set it before starting:
export AILOOT_API_KEY="sk-..." # add to ~/.zshrc or ~/.bashrc
setx AILOOT_API_KEY "sk-..." # then restart the terminal
Example model: ailoot/claude-sonnet-5-5, ailoot/gpt-6.1-sol
Important
Set url to the base URL with /v1, not the full /chat/completions path. Chat Completions is the default; api: "responses" switches to the Responses API. The id form "provider/model" sends the bare model ID to the gateway. The documentation shows apiKey in the model object in an example with headers; its use alongside url is not described separately.
AsyncOpenAI(base_url=..., api_key=...) + OpenAIChatCompletionsModel(model=..., openai_client=client), or set_default_openai_client(client) globally
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, POST /v1/responses, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
import os
from agents import Agent, Runner, AsyncOpenAI, OpenAIChatCompletionsModel, set_tracing_disabled
set_tracing_disabled(disabled=True)
client = AsyncOpenAI(max_retries=0, base_url="https://api.ailoot.ai/v1", api_key=os.environ["AILOOT_API_KEY"])
model = OpenAIChatCompletionsModel(model="claude-sonnet-5-5", openai_client=client)
agent = Agent(name="Assistant", instructions="Respond in Russian", model=model)
print(Runner.run_sync(agent, "Hello").final_output)
Example model: claude-sonnet-5-5, gpt-6.1-sol
Important
Use an address with /v1. Disable tracing with set_tracing_disabled(True): otherwise, the SDK will attempt to send traces to OpenAI without an OpenAI key. Chat Completions is available only if the model lists this method in the catalog and the method is available to your key; Responses (the SDK default) is available only for models that list the responses method in the catalog. OpenAI’s built-in tools (web search and files) are not guaranteed through a third-party gateway.
OpenAIProvider(base_url=..., api_key=...) with OpenAIChatModel for Chat Completions or OpenAIResponsesModel for Responses, or the OPENAI_BASE_URL variable
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, POST /v1/responses, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
import os
from pydantic_ai import Agent
from pydantic_ai.models.openai import OpenAIChatModel
from pydantic_ai.providers.openai import OpenAIProvider
model = OpenAIChatModel("deepseek-v4-pro",
provider=OpenAIProvider(base_url="https://api.ailoot.ai/v1", api_key=os.environ["AILOOT_API_KEY"]))
agent = Agent(model)
print(agent.run_sync("Hello").output)
Example model: deepseek-v4-pro, gpt-6.1-sol (for OpenAIResponsesModel)
Important
Include /v1 in the URL. The documentation defaults to OpenAIResponsesModel: use it only for models with the responses method; use OpenAIChatModel for the others.
Credentials → OpenAI → Base URL (default https://api.openai.com/v1); Credentials → Anthropic → Base URL (default https://api.anthropic.com); Credentials → Google Gemini(PaLM) Api → Host (default https://generativelanguage.googleapis.com). Then use OpenAI Chat Model / Anthropic Chat Model / Google Gemini Chat Model nodes
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, POST /v1/responses, POST /v1/messages, POST /v1beta/models/{model}:generateContent, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
OpenAI credential:
API Key: sk-...
Base URL: https://api.ailoot.ai/v1
Anthropic credential:
API Key: sk-...
Base URL: https://api.ailoot.ai
Google Gemini(PaLM) Api credential:
Host: https://api.ailoot.ai
API Key: sk-...
OpenAI Chat Model node → Model: By ID → gpt-6.1-sol (Use Responses API toggle — only for models with the responses method)
Example model: gpt-6.1-sol / deepseek-v4-pro (OpenAI), claude-sonnet-5-5 (Anthropic), gemini-3.6-flash (Gemini)
Important
Credential checks call /models for OpenAI, /v1/models for Anthropic, and /v1beta/models for Gemini; Ailoot provides all three paths. The n8n documentation pages do not describe Base URL, but the field exists in the credentials themselves: OpenAiApi.credentials.ts and AnthropicApi.credentials.ts label it “Override the default base URL for the API”. On n8n Cloud, do not select “Use Gateway credits”, or requests will bypass Ailoot.
Integrations → Model Provider → install the OpenAI-API-compatible plugin → Add Model (fields: Model Type, Model Name, API Key, API Base URL, Completion mode, Model context size)
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
Model Type: LLM
Model Name: claude-sonnet-5-5
API Key: sk-...
API Base URL: https://api.ailoot.ai/v1
Completion mode: Chat
Model context size: as listed on the model card in the Ailoot catalog
For Text Embedding:
Model Name: nemotron-3-embed
API Base URL: https://api.ailoot.ai
Example model: claude-sonnet-5-5, gpt-6.1-sol, qwen3.8-max; nemotron-3-embed (embeddings)
Important
Only the workspace owner and administrators can manage providers. According to the plugin README, the URL includes /v1 for LLMs; for embeddings, rerank, STT, and TTS, the plugin appends the API version, so use a URL without /v1 (otherwise the result will be /v1/v1). Add each Ailoot model as a separate entry; enable image input and streaming output in the model settings.
ChatOpenAI or ChatOpenAI Custom node → Connect Credential (key) → Additional Parameters → Base Path (and Base Options for headers)
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
Credential (OpenAI API): sk-...
ChatOpenAI Custom node:
Model Name: deepseek-v4-pro
Additional Parameters → Base Path: https://api.ailoot.ai/v1
Example model: deepseek-v4-pro, claude-sonnet-5-5, gpt-6.1-sol
Important
ChatOpenAI has a fixed model list, so use ChatOpenAI Custom for arbitrary Ailoot ids; it lets you enter the model name manually. Base Path includes /v1. For embeddings, use the OpenAI Embeddings Custom node.
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
API Key: sk-...
Base Url: https://api.ailoot.ai/v1
Options → Chat Model: gpt-6.1-sol
Voice Assistants → Conversation agent: Extended OpenAI Conversation
Example model: gpt-6.1-sol, claude-sonnet-5-5
Important
This is a third-party component, not an official integration. Device control uses function calling (tools) in Chat Completions, so choose models that support tools. The official OpenAI Conversation integration does not support a custom endpoint.
Profile → Settings → Model Providers → OpenAI Compatible (Base URL and API Key fields; OPENAI_COMPATIBLE_BASE_URL and OPENAI_COMPATIBLE_API_KEY environment variables). Alternatively: OpenAI component → OpenAI API Base field
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
Settings → Model Providers → OpenAI Compatible:
Base URL: https://api.ailoot.ai/v1
API Key: sk-...
or through environment variables:
export OPENAI_COMPATIBLE_BASE_URL="https://api.ailoot.ai/v1"
export OPENAI_COMPATIBLE_API_KEY="sk-..."
Then select OpenAI Compatible → claude-sonnet-5-5 in the Language Model / Agent components
Example model: claude-sonnet-5-5, gpt-6.1-sol, nemotron-3-embed (Embedding Model)
Important
Langflow tests the connection and fetches the model list from /v1/models. The models appear under Language Models and Embedding Models; you then need to enable the models you want manually. The provider is installed with langflow.
The built-in OpenAI module works only with an OpenAI account. Use HTTP → Make a request: URL, Method, Headers, Body type Raw / JSON
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
Module: HTTP → Make a request
URL: https://api.ailoot.ai/v1/chat/completions
Method: POST
Headers:
Authorization: Bearer sk-...
Content-Type: application/json
Body (JSON):
{"model":"gpt-6.1-sol","messages":[{"role":"user","content":"{{1.text}}"}]}
Parse response: Yes → the response is in choices[].message.content
Example model: gpt-6.1-sol, claude-sonnet-5-5
Important
This is a general-purpose HTTP module, not a dedicated Ailoot provider: its documentation allows any HTTPS URL. Store the key in an API-key connection, as Make recommends, rather than passing it in headers. Make does not need streaming output; omit stream.
Telegram frameworks have no built-in LLM layer: the message handler calls the OpenAI SDK with the Ailoot base_url, or the Anthropic SDK with base_url without /v1
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
Save the Python code as ailoot-telegram-bot.py. Install the dependencies, set both environment variables, then run the file. Send Hello to your bot and check for a text reply. If the request fails, keep its Request ID and check the original result before sending another generation.
python3 -m pip install "aiogram>=3,<4" openai
ailoot-telegram-bot.py
import asyncio
import os
from aiogram import Bot, Dispatcher, F
from aiogram.types import Message
from openai import AsyncOpenAI, APIError
async def main():
dp = Dispatcher()
async with AsyncOpenAI(
base_url="https://api.ailoot.ai/v1",
api_key=os.environ["AILOOT_API_KEY"],
max_retries=0,
) as llm:
@dp.message(F.text)
async def reply(message: Message):
try:
result = await llm.chat.completions.create(
model="claude-sonnet-5-5",
messages=[{"role": "user", "content": message.text}],
)
content = result.choices[0].message.content or "—"
for start in range(0, len(content), 4000):
await message.answer(content[start:start + 4000])
except APIError as error:
await message.answer(
f"{type(error).__name__}; Request ID: "
f"{getattr(error, 'request_id', None) or '—'}"
)
bot = Bot(token=os.environ["TELEGRAM_BOT_TOKEN"])
await dp.start_polling(bot)
if __name__ == "__main__":
asyncio.run(main())
Example model: claude-sonnet-5-5, deepseek-v4-pro, qwen3.8-flash
Important
Telegram frameworks have no built-in AI layer: configure the Ailoot URL in the OpenAI SDK inside the bot. With a zero balance, Ailoot returns a clear 402 error with a Request ID; show it to the bot user instead of crashing.
Zapier’s ChatGPT (OpenAI) application works only with OpenAI. Use Webhooks by Zapier → Custom Request (Method, URL, Data, Headers)
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
Action: Webhooks by Zapier → Custom Request
Method: POST
URL: https://api.ailoot.ai/v1/chat/completions
Data (raw JSON):
{"model":"claude-sonnet-5-5","messages":[{"role":"user","content":"Hello"}]}
Headers:
Authorization | Bearer sk-...
Content-Type | application/json
Example model: claude-sonnet-5-5, gpt-6.1-sol
Important
This is a general-purpose HTTP request. Zapier has no ready-made provider with a base URL field. Custom Request sends Data unchanged without parsing it, so validate the JSON beforehand. Webhooks by Zapier is available on paid Zapier plans.
~/.gemini/antigravity-cli/settings.json → "modelProvider": "gemini"; GEMINI_API_KEY and GOOGLE_GEMINI_BASE_URL environment variables
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1beta/models/{model}:generateContent, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
~/.gemini/antigravity-cli/settings.json
{
"modelProvider": "gemini"
}
export AILOOT_API_KEY="sk-..."
# in the terminal:
export GEMINI_API_KEY="$AILOOT_API_KEY"
export GOOGLE_GEMINI_BASE_URL=https://api.ailoot.ai
agy --model gemini-3.8-flash
Example model: gemini-3.8-flash
Important
The protocol is native Gemini REST (streamGenerateContent, key in X-Goog-Api-Key), using the https://api.ailoot.ai/v1beta/... path. For each question, agy sends two requests: the conversation title and the response itself. The agent reserves substantially more for each step than a regular chat, so keep extra funds in your balance. This card is in the “Under review” section: the configuration is based on the documentation, and a live client run through Ailoot has not yet been performed. If modelProvider = gemini and GEMINI_API_KEY is empty, the CLI will not start.
cline auth → custom API providers → OpenAI Compatible: Base URL, API Key, Model ID; the -P/--provider flag selects the provider
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
cline auth
# API Provider: OpenAI Compatible
# Base URL: https://api.ailoot.ai/v1
# API Key: sk-...
# Model ID: claude-sonnet-5-5
cline "Run the tests and fix the errors"
Example model: claude-sonnet-5-5
Important
The Base URL / API Key / Model ID fields are described in Cline's documentation for OpenAI Compatible. The CLI documentation does not list the exact labels in the cline auth wizard, so check the screen.
GitHub confirms custom endpoint support. Its official screenshot shows Provider Name, API Key and API Type: Chat Completions, but no endpoint URL example. A ready Ailoot configuration is therefore not provided here. JetBrains AI Assistant is the documented alternative in this IDE. Ask the official plugin support team to clarify the endpoint URL requirements.
Cursor Settings → Models → API Keys → OpenAI API Key + “Override OpenAI Base URL” toggle; then add the model ID to your custom models.
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
OpenAI API Key: sk-...
Override OpenAI Base URL: https://api.ailoot.ai/v1
Custom model: gpt-6.1-sol (add, enable, select in chat)
Example model: gpt-6.1-sol
Important
Cursor's current official documentation does not describe a custom endpoint: it lists only OpenAI/Anthropic/Google/Azure/Bedrock keys, states that custom keys work only in chat (Tab continues to use Cursor models), and says that all requests go through Cursor's servers (the endpoint must be public; api.ailoot.ai is public). The URL override toggle exists in the interface, but Cursor forum users report issues: it applies to all models, Agent mode may not use a custom key, and the field does not accept input in some 3.15.x builds.
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
Provider: OpenAI Chat
API Key: sk-...
Base Path: https://api.ailoot.ai/v1
Model: gpt-6.1-sol
Example model: gpt-6.1-sol
Important
The plugin documentation mentions a Custom provider but does not describe the URL field or its format. These settings are based on the plugin interface; check your settings window.
Cmd/Ctrl+Shift+P → Open config.json → an entry in models[] (PearAI is a Continue fork).
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
The official PearAI documentation could not be checked because the documentation site was unavailable, and the repository has not been updated for a long time. The configuration follows Continue’s example; check it in the application.
Chat → model picker → Manage models → Add custom model → OpenAI (custom endpoint) → URL, key, and model.
Configuration
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
The Visual Studio blog (29.09.2026) announced custom URLs for OpenAI and Ollama in VS 2026 Stable. The documentation still describes only OpenAI/Anthropic/Google keys without a URL field, so field names and URL format remain unconfirmed. Chat only; unavailable on Copilot Business/Enterprise plans.
Before starting, request GET /v1/models with your Ailoot key. Choose an exact model ID whose methods include POST /v1/chat/completions, then use that ID in the configuration. For agents, also check tool support and the features required by your client version.
These settings come from Void’s source code, not its user documentation. The Void repository is archived, and the editor is no longer maintained; this is useful only for existing users.
No results found. If the application supports a custom OpenAI or Anthropic API URL, it will most likely work. Here is what to enter.
Not currently compatible
These applications do not support a custom API URL. Alternatives are listed alongside them.
Application
Why it is not compatible
What to use instead
Cursor CLI (cursor-agent)
Uses only Cursor models; you cannot set a custom API URL.
Claude Code, Codex CLI, or opencode
Auggie CLI (Augment Code)
Requires sign-in with an Augment account; you cannot specify a third-party provider.
Claude Code or opencode
Windsurf / Devin Desktop
Accepts only an Anthropic key for selected Claude models; you cannot set a custom URL.
Cline, Roo Code, or Kilo Code in VS Code
Kiro
Does not support your own key or a custom API URL.
Cline or Roo Code
Google Antigravity (IDE)
There is no official setting for a custom provider; we do not recommend third-party patches.
Antigravity CLI or Gemini CLI
Visual Studio with Copilot Business or Enterprise
Custom models are not available on these Copilot plans.
An individual Copilot plan — see the Visual Studio 2026 integration card
Cursor Tab autocomplete
Always uses Cursor models, even when you supply your own key.
Cursor chat is configured separately — see the Cursor integration card
Enchanted
Works only with the Ollama API.
Open WebUI or Chatbox
Smart Connections (Obsidian)
External APIs are available only with paid Smart Chat Pro.
Obsidian Copilot
Witsy, 5ire
We could not find official documentation for setting a custom API URL.
Cherry Studio or Jan
Make: built-in OpenAI module
Connects only to an OpenAI account.
HTTP module → Make a request — see the Make integration card
Zapier: ChatGPT (OpenAI) app
There is no field for a custom URL.
Webhooks by Zapier → Custom Request — see the Zapier integration card
Home Assistant: OpenAI Conversation
The built-in integration works only with the OpenAI URL.
Extended OpenAI Conversation from HACS — see the Home Assistant integration card
aiogram, python-telegram-bot, grammY, Telegraf
Bot frameworks do not connect to models on their own.
OpenAI or Anthropic SDK inside your bot — see the Telegram bot integration card
More details
AI API connections and client setup
How do I set the AI API URL and key in my client?
Agents, editors, chats, and SDKs that accept an Ailoot key. The “Set up” button shows where to enter the URL and key, with a ready-to-copy configuration.
Connect Ailoot to Claude Code, OpenCode, Hermes or your own app.
The company and its project portfolio
The company and its project portfolio: millions of customers; monthly sales in the billions.
Buy tokens at a discount.
Subscription discounts for regular use. Promo codes offer up to 50% off token purchases.
Subscription
Choose a discounted plan for your regular requests.
Promo code: up to 50% off
Enter your code at checkout. The price updates to reflect the current promotion.
Spending in your history
The model, tokens and amount are recorded for each request.
Choose the OpenAI, Anthropic or Gemini format supported by your client. Check the Ailoot model card for a compatible API method and model ID; one supported format does not mean every model supports all formats.
What do I need to configure in my client?
Set the Ailoot API key, API address and a model ID compatible with the client method. OpenAI SDK uses https://api.ailoot.ai/v1; Anthropic and Gemini SDK use https://api.ailoot.ai. Keep the request fields from the example for that format.
How do I send my first API request?
Create an Ailoot API key. Set https://api.ailoot.ai/v1 for OpenAI SDK, or https://api.ailoot.ai for Anthropic and Gemini SDK. GET /v1/models is free and lists models available to your key; then use the example for the selected model and API method.
What should I check when a connected client fails?
Check that the client sends the expected OpenAI, Anthropic or Gemini format to the matching API address with your Ailoot API key. Confirm the model ID and method, then compare the returned error with the client guide. Send the error and request ID to support@ailoot.ai without sharing your key.
Can I use Ailoot in a paid product?
Yes. Connect the API to your SaaS, app or bot under the Ailoot service offer and licence agreement. For resale, request written terms and a reseller licence for your project.
How do I request a reseller licence?
Email support@ailoot.ai with what your product does, who uses it, and how you provide access. Request a reseller licence and written terms for your project.