Qwen3.5 highlights: the 122B-A10B expert model
Qwen3.5-122B-A10B is a post-trained vision-language model for joint text and visual tasks. It accepts text, images and video and returns text, including explanations, code proposals and requested document artifacts. The family is built through early text–vision fusion and reinforcement learning on progressively complex tasks, developing reasoning, coding, agents and visual understanding. It describes coverage of 201 languages and dialects, with no implication that every language or domain has identical reliability.
A useful task can combine a technical specification, diagrams and relevant code, asking the model to explain the components and return an implementation proposal. A document-analysis task can ask for a comparison with supporting section or page references.
Model overview
The sparse mixture-of-experts checkpoint has 122 billion parameters in total and activates ten billion. It has 48 language-model layers, with groups of three Gated DeltaNet expert layers followed by one gated-attention expert layer. The maker lists 256 routed experts, selecting eight routed experts together with one shared expert. A vision encoder and multi-token prediction are part of the model. Its total and activated counts have different meanings; the activated figure is not the complete checkpoint size.
This is a distinct edition from dense Qwen3.5-27B, sparse Qwen3.5-35B-A3B and later model generations. Do not infer a general performance or latency ranking from the size alone. The native context is 262,144 tokens and the documented extension reaches 1,010,000 with configuration. Qwen advises retaining at least a 128K context where possible for complex thinking, while recognizing that memory pressure may require a smaller window.
Preparing joint text and visual evidence
Preserve diagram labels, chart axes, table headings and units. State whether the result should explain a relationship, extract named fields or propose code. When several sources matter, label them and request an output organized by the same identifiers. Separate visible facts from assumptions so a reviewer can trace the reasoning to the supplied material.
For a coding request, identify the intended behavior, relevant source and an input that demonstrates the issue. Ask for an editable patch with an explanation of the affected conditions and checks. For video, preserve the event sequence and timing and specify the question rather than request an undifferentiated summary. The model’s text result is a draft artifact to inspect; visual input capability does not mean image, video or speech generation.
Thinking and history
Thinking is enabled by default, with reasoning before the final answer. The maker’s enable_thinking setting can disable it for direct responses. The old /think and /nothink soft switches are not officially supported. Qwen3.5 recommends keeping final answers without thinking content in historical assistant messages. This should not be replaced by later Qwen generations’ historical preservation options simply because they share a family name.
Keep the final result easy to locate, such as a defined answer field, a delimited value or a separate code artifact. A direct answer still needs checking, while longer reasoning is useful only when its steps support a verifiable conclusion.
Agentic usage
Qwen documents tool calling and recommends Qwen-Agent with configured or MCP-connected tools, along with Qwen Code for coding-agent work. Define the permitted actions, available evidence and completion criteria. Connected tools execute operations and return results; the model uses that feedback in its analysis. A proposed command, retrieval query or patch should remain distinct from an action actually completed by the application.
Ultra-long context and video
The maker’s extension uses YaRN with the checkpoint’s interleaved positional settings. A factor of four targets the roughly one-million-token workflow, and two is suggested for 524,288-token workloads. Static scaling can affect shorter-text performance, so it should follow a real need for larger context. Input, represented visuals, history, reasoning and the final response all use the configured sequence budget.
The supported video example defaults to two frames per second. For hour-scale material, the maker recommends increasing the processor budget to roughly 224K video tokens for higher-rate sampling. That is a configuration-dependent recipe, not a universal media-duration, complete-recall or timestamp-precision guarantee.
Generation guidance and substantive limits
Best Practices recommend temperature 1.0, top_p 0.95, top_k 20, min_p 0, presence penalty 1.5 and repetition penalty 1.0 for general thinking. Precise coding uses temperature 0.6 and presence penalty zero. General direct responses use temperature 0.7, top_p 0.8 and presence penalty 1.5; top_k is 20, min_p is 0 and repetition penalty is 1.0. Direct-response reasoning has a separate preset: temperature 1.0, top_p 1.0, top_k 40 and presence penalty 2.0. Higher presence penalties can mix languages or slightly reduce quality.
Recommended generation allowances are 32,768 tokens for most queries and 81,920 for unusually complex mathematical or programming work; they are not separately established hard output ceilings. Define the requested format, retain source identifiers and check calculations, extracted values and proposed code against the original evidence. The useful outcome is a complete textual artifact whose assumptions and supporting material remain visible.
", "aboutText": "Qwen3.5 highlights: the 122B-A10B expert model\n\nQwen3.5-122B-A10B is a post-trained vision-language model for joint text and visual tasks. It accepts text, images and video and returns text, including explanations, code proposals and requested document artifacts. The family is built through early text–vision fusion and reinforcement learning on progressively complex tasks, developing reasoning, coding, agents and visual understanding. It describes coverage of 201 languages and dialects, with no implication that every language or domain has identical reliability.\n\nA useful task can combine a technical specification, diagrams and relevant code, asking the model to explain the components and return an implementation proposal. A document-analysis task can ask for a comparison with supporting section or page references.\n\nModel overview\n\nThe sparse mixture-of-experts checkpoint has 122 billion parameters in total and activates ten billion. It has 48 language-model layers, with groups of three Gated DeltaNet expert layers followed by one gated-attention expert layer. The maker lists 256 routed experts, selecting eight routed experts together with one shared expert. A vision encoder and multi-token prediction are part of the model. Its total and activated counts have different meanings; the activated figure is not the complete checkpoint size.\n\nThis is a distinct edition from dense Qwen3.5-27B, sparse Qwen3.5-35B-A3B and later model generations. Do not infer a general performance or latency ranking from the size alone. The native context is 262,144 tokens and the documented extension reaches 1,010,000 with configuration. Qwen advises retaining at least a 128K context where possible for complex thinking, while recognizing that memory pressure may require a smaller window.\n\nPreparing joint text and visual evidence\n\nPreserve diagram labels, chart axes, table headings and units. State whether the result should explain a relationship, extract named fields or propose code. When several sources matter, label them and request an output organized by the same identifiers. Separate visible facts from assumptions so a reviewer can trace the reasoning to the supplied material.\n\nFor a coding request, identify the intended behavior, relevant source and an input that demonstrates the issue. Ask for an editable patch with an explanation of the affected conditions and checks. For video, preserve the event sequence and timing and specify the question rather than request an undifferentiated summary. The model’s text result is a draft artifact to inspect; visual input capability does not mean image, video or speech generation.\n\nThinking and history\n\nThinking is enabled by default, with reasoning before the final answer. The maker’s enable_thinking setting can disable it for direct responses. The old /think and /nothink soft switches are not officially supported. Qwen3.5 recommends keeping final answers without thinking content in historical assistant messages. This should not be replaced by later Qwen generations’ historical preservation options simply because they share a family name.\n\nKeep the final result easy to locate, such as a defined answer field, a delimited value or a separate code artifact. A direct answer still needs checking, while longer reasoning is useful only when its steps support a verifiable conclusion.\n\nAgentic usage\n\nQwen documents tool calling and recommends Qwen-Agent with configured or MCP-connected tools, along with Qwen Code for coding-agent work. Define the permitted actions, available evidence and completion criteria. Connected tools execute operations and return results; the model uses that feedback in its analysis. A proposed command, retrieval query or patch should remain distinct from an action actually completed by the application.\n\nUltra-long context and video\n\nThe maker’s extension uses YaRN with the checkpoint’s interleaved positional settings. A factor of four targets the roughly one-million-token workflow, and two is suggested for 524,288-token workloads. Static scaling can affect shorter-text performance, so it should follow a real need for larger context. Input, represented visuals, history, reasoning and the final response all use the configured sequence budget.\n\nThe supported video example defaults to two frames per second. For hour-scale material, the maker recommends increasing the processor budget to roughly 224K video tokens for higher-rate sampling. That is a configuration-dependent recipe, not a universal media-duration, complete-recall or timestamp-precision guarantee.\n\nGeneration guidance and substantive limits\n\nBest Practices recommend temperature 1.0, top_p 0.95, top_k 20, min_p 0, presence penalty 1.5 and repetition penalty 1.0 for general thinking. Precise coding uses temperature 0.6 and presence penalty zero. General direct responses use temperature 0.7, top_p 0.8 and presence penalty 1.5; top_k is 20, min_p is 0 and repetition penalty is 1.0. Direct-response reasoning has a separate preset: temperature 1.0, top_p 1.0, top_k 40 and presence penalty 2.0. Higher presence penalties can mix languages or slightly reduce quality.\n\nRecommended generation allowances are 32,768 tokens for most queries and 81,920 for unusually complex mathematical or programming work; they are not separately established hard output ceilings. Define the requested format, retain source identifiers and check calculations, extracted values and proposed code against the original evidence. The useful outcome is a complete textual artifact whose assumptions and supporting material remain visible.", "editorialOverlay": { "copy_fields": { "practical_html": "Technical example in English
Check a process graph against its rule\n\nA process specification says validation must complete before persistence. Textual graph G1 is receive -> persist -> validate -> acknowledge. Code excerpt: save(record); validate(record); return "accepted". Propose an implementation correction and update the graph in text. Include a test showing that an invalid record never reaches save.A corrected sequence diagram in text, a focused code proposal and a test using an invalid record.Check that the graph and code agree on validate-before-save and that acknowledgment follows successful persistence.qwen/qwen3.5-122b-a10bQwen3.5-122B-A10B is Qwen’s vision-language mixture-of-experts model with 122 billion total and 10 billion activated parameters. It takes text, images and video for reasoning, coding and agent tasks, returning text with selectable thinking and a native 262,144-token context.
This description covers the model’s capabilities. Available in Ailoot inputs, outputs and options are listed for the selected method below.
Qwen3.5 highlights: the 122B-A10B expert model
Qwen3.5-122B-A10B is a post-trained vision-language model for joint text and visual tasks. It accepts text, images and video and returns text, including explanations, code proposals and requested document artifacts. The family is built through early text–vision fusion and reinforcement learning on progressively complex tasks, developing reasoning, coding, agents and visual understanding. It describes coverage of 201 languages and dialects, with no implication that every language or domain has identical reliability.
A useful task can combine a technical specification, diagrams and relevant code, asking the model to explain the components and return an implementation proposal. A document-analysis task can ask for a comparison with supporting section or page references.
Model overview
The sparse mixture-of-experts checkpoint has 122 billion parameters in total and activates ten billion. It has 48 language-model layers, with groups of three Gated DeltaNet expert layers followed by one gated-attention expert layer. The maker lists 256 routed experts, selecting eight routed experts together with one shared expert. A vision encoder and multi-token prediction are part of the model. Its total and activated counts have different meanings; the activated figure is not the complete checkpoint size.
This is a distinct edition from dense Qwen3.5-27B, sparse Qwen3.5-35B-A3B and later model generations. Do not infer a general performance or latency ranking from the size alone. The native context is 262,144 tokens and the documented extension reaches 1,010,000 with configuration. Qwen advises retaining at least a 128K context where possible for complex thinking, while recognizing that memory pressure may require a smaller window.
Preparing joint text and visual evidence
Preserve diagram labels, chart axes, table headings and units. State whether the result should explain a relationship, extract named fields or propose code. When several sources matter, label them and request an output organized by the same identifiers. Separate visible facts from assumptions so a reviewer can trace the reasoning to the supplied material.
For a coding request, identify the intended behavior, relevant source and an input that demonstrates the issue. Ask for an editable patch with an explanation of the affected conditions and checks. For video, preserve the event sequence and timing and specify the question rather than request an undifferentiated summary. The model’s text result is a draft artifact to inspect; visual input capability does not mean image, video or speech generation.
Thinking and history
Thinking is enabled by default, with reasoning before the final answer. The maker’s enable_thinking setting can disable it for direct responses. The old /think and /nothink soft switches are not officially supported. Qwen3.5 recommends keeping final answers without thinking content in historical assistant messages. This should not be replaced by later Qwen generations’ historical preservation options simply because they share a family name.
Keep the final result easy to locate, such as a defined answer field, a delimited value or a separate code artifact. A direct answer still needs checking, while longer reasoning is useful only when its steps support a verifiable conclusion.
Agentic usage
Qwen documents tool calling and recommends Qwen-Agent with configured or MCP-connected tools, along with Qwen Code for coding-agent work. Define the permitted actions, available evidence and completion criteria. Connected tools execute operations and return results; the model uses that feedback in its analysis. A proposed command, retrieval query or patch should remain distinct from an action actually completed by the application.
Ultra-long context and video
The maker’s extension uses YaRN with the checkpoint’s interleaved positional settings. A factor of four targets the roughly one-million-token workflow, and two is suggested for 524,288-token workloads. Static scaling can affect shorter-text performance, so it should follow a real need for larger context. Input, represented visuals, history, reasoning and the final response all use the configured sequence budget.
The supported video example defaults to two frames per second. For hour-scale material, the maker recommends increasing the processor budget to roughly 224K video tokens for higher-rate sampling. That is a configuration-dependent recipe, not a universal media-duration, complete-recall or timestamp-precision guarantee.
Generation guidance and substantive limits
Best Practices recommend temperature 1.0, top_p 0.95, top_k 20, min_p 0, presence penalty 1.5 and repetition penalty 1.0 for general thinking. Precise coding uses temperature 0.6 and presence penalty zero. General direct responses use temperature 0.7, top_p 0.8 and presence penalty 1.5; top_k is 20, min_p is 0 and repetition penalty is 1.0. Direct-response reasoning has a separate preset: temperature 1.0, top_p 1.0, top_k 40 and presence penalty 2.0. Higher presence penalties can mix languages or slightly reduce quality.
Recommended generation allowances are 32,768 tokens for most queries and 81,920 for unusually complex mathematical or programming work; they are not separately established hard output ceilings. Define the requested format, retain source identifiers and check calculations, extracted values and proposed code against the original evidence. The useful outcome is a complete textual artifact whose assumptions and supporting material remain visible.
POST /v1/chat/completionstexttext| Field | Type | Requirement | Values & limits | Default |
|---|---|---|---|---|
Fieldmodel | Typestring | RequirementRequired | Values & limits qwen/qwen3.5-122b-a10b | Default— |
Fieldmessages | Typemessage[] | RequirementRequired | Values & limits Minimum items: 1 | Default— |
Fieldmessages[].role | Typestring | RequirementRequired | Values & limits Examples: system, user, assistant, tool, function Conditions: These roles are explicitly represented by Source compatibility types; the exact model role set remains model-specific. Model-specific | Default— |
Fieldmessages[].content | Typestring|content_part[]|null | RequirementOptional | Values & limits | Default— |
Fieldmessages[].tool_calls | Typetool_call[] | RequirementOptional | Values & limits | Default— |
Fieldmessages[].tool_call_id | Typestring | RequirementOptional | Values & limits Applies when: role=tool; use actual returned ID | Default— |
Fieldmessages[].name | Typestring | RequirementOptional | Values & limits | Default— |
Fieldmessages[].reasoning_content | Typestring | RequirementOptional | Values & limits Applies when: Round-trip actual returned reasoning where this model uses it | Default— |
Fieldmessages[].reasoning_details | Typeobject|array | RequirementOptional | Values & limits Model-specific | Default— |
Fieldmax_tokens | Typeinteger | RequirementOptional | Values & limits Minimum: 1 Model-specific | Default— |
Fieldmax_completion_tokens | Typeinteger | RequirementOptional | Values & limits Minimum: 1 Model-specific | Default— |
Fieldstream | Typeboolean | RequirementOptional | Values & limits | Defaultfalse |
Fieldstream_options.include_usage | Typeboolean | RequirementOptional | Values & limits Model-specific | Default— |
Fieldtemperature | Typenumber | RequirementOptional | Values & limits Model-specific | Default— |
Fieldtop_p | Typenumber | RequirementOptional | Values & limits Model-specific | Default— |
Fieldstop | Typestring|string[] | RequirementOptional | Values & limits Model-specific | Default— |
Fieldtools | Typefunction_tool[] | RequirementOptional | Values & limits Model-specific | Default— |
Fieldtools[].type | Typestring | RequirementOptional | Values & limits function Model-specific | Default— |
Fieldtools[].function.name | Typestring | RequirementRequired | Values & limits Applies when: function tool supplied | Default— |
Fieldtools[].function.description | Typestring | RequirementOptional | Values & limits | Default— |
Fieldtools[].function.parameters | TypeJSON_schema | RequirementOptional | Values & limits | Default— |
Fieldtool_choice | Typestring|object | RequirementOptional | Values & limits Model-specific | Default— |
Fieldparallel_tool_calls | Typeboolean | RequirementOptional | Values & limits Model-specific | Default— |
Fieldresponse_format | Typeobject | RequirementOptional | Values & limits Model-specific | Default— |
Fieldreasoning_effort | Typestring | RequirementOptional | Values & limits Model-specific | Default— |
Fieldreasoning | Typeobject | RequirementOptional | Values & limits Model-specific | Default— |
Fieldservice_tier | Typestring | RequirementOptional | Values & limits Model-specific | Default— |
Fieldinstructions | Typestring | RequirementOptional | Values & limits Compatibility field | Default— |
Fieldfunctions | Typefunction[] | RequirementOptional | Values & limits Model-specific | Default— |
Fieldfunction_call | Typestring|object | RequirementOptional | Values & limits Model-specific | Default— |
Fieldprompt_cache_options | Typeobject | RequirementOptional | Values & limits Model-specific | Default— |
Fieldmodalities | Typestring[] | RequirementOptional | Values & limits text | Default— |
Fieldprovider | Typeobject | RequirementOptional | Values & limits Conditions: provider.options is rejected on this Ultra text route | Default— |
Fieldcache_control.ttl | Typestring | RequirementOptional | Values & limits Conditions: 1h is rejected in actual prompt-cache locations | Default— |
Text · UTF8 text
messages[].content
| Parameter | Contract |
|---|---|
Parametercontent_type | Contractapplication/json |
Parameterrequired | Contractmodel, messages |
Parametermaterial_types | Contracttext |
Parametermessages | Contractmessage array with role/content shape |
Parameterfile_material | Contractactual UTF-8 document inserted as literal text; no file ID inferred |
Parameteroptional_controls | Contractstream, temperature, top_p, stop, max_tokens, max_completion_tokens, reasoning, response_format, tools, tool_choice |
Parametercontrol_scope | Contractmodel-specific; user function tools differ from hosted tools |
Parameteroutput_modalities | Contracttext |
Parameterhosted_operations | Contractfalse |
Parametermultimodal_input_binding | Contractnull |
Lifecycle: synchronous_or_streamed
Send exact model ID with messages.
Context length is not an output cap.
A text file may supply literal text; image/audio/file-object inputs need exact binding.
Tariff references do not add request options. Supported parameters are listed for the selected method.
Match each charge to the conditions shown. Use one rate for each set of alternative configurations.
| Configuration | Rate schedule | Context tokens, more than | Context tokens, up to and including | Charge |
|---|---|---|---|---|
| ConfigurationBase token rates | Rate scheduleBase rate | Context tokens, more than— | Context tokens, up to and includingNo upper bound | Charge Input · Input tokens $0.00058 / 1K tokens Output · Output tokens $0.00458 / 1K tokens Input · Cache read $0 / 1K tokens Input · Cache write $0 / 1K tokens Input · Cache write, 1 hour — / 1K tokens |
| ConfigurationRequest-length tier 1 | Rate schedule— | Context tokens, more than0 | Context tokens, up to and includingNo upper bound | Charge Input · Input tokens $0.00058 / 1K tokens Output · Output tokens $0.00458 / 1K tokens Input · Cache read $0 / 1K tokens Input · Cache write $0 / 1K tokens Input · Cache write, 1 hour — / 1K tokens |
Ailoot reserves an estimated amount before a request. After the result, it settles the final charge from measured usage. Streaming settlement uses the final usage record. How reservation and settlement work →
set -eu\n: "${AILOOT_API_KEY:?Set your Ailoot API key}"\npython3 - <<'PY_REQUEST'\nimport json,os\nfrom pathlib import Path\np=json.loads('{"model": "qwen/qwen3.5-122b-a10b", "messages": [{"role": "user", "content": "Explain the difference between an observation and a hypothesis in three sentences."}]}')\n\nPath("request.json").write_text(json.dumps(p,ensure_ascii=False))\nPY_REQUEST\ncurl --fail-with-body --silent --show-error https://api.ailoot.ai/v1/chat/completions \\\n -H "Authorization: Bearer $AILOOT_API_KEY" -H 'Content-Type: application/json' \\\n --data-binary @request.json -D response-headers.txt -o result.json\npython3 - <<'PY_RESULT'\nimport json\nfrom pathlib import Path\nr=json.loads(Path("result.json").read_text())\nfor x in r.get("choices",[]):\n if x.get("message",{}).get("content"): print(x["message"]["content"])\nPY_RESULT\nfrom openai import OpenAI\nimport os\n\nif not os.environ.get("AILOOT_API_KEY", "").strip(): raise SystemExit("Set AILOOT_API_KEY before making a request")\nclient=OpenAI(base_url="https://api.ailoot.ai/v1",api_key=os.environ["AILOOT_API_KEY"],max_retries=0)\nresult=client.chat.completions.create(model='qwen/qwen3.5-122b-a10b',messages=[{"role":"user","content":"Explain observations and hypotheses in three sentences."}])\nimport base64, json\nfrom pathlib import Path\n\ndef save_result(data, stem="result"):\n # Keep every field, including usage, IDs, reasoning and tool requests.\n Path(stem + ".json").write_text(json.dumps(data, ensure_ascii=False, indent=2), encoding="utf-8")\n print(json.dumps(data, ensure_ascii=False, indent=2))\n def save_blob(encoded, suffix, label):\n raw = base64.b64decode(encoded, altchars=b"-_", validate=True)\n ext = "png" if raw.startswith(b"\\x89PNG\\r\\n\\x1a\\n") else "jpg" if raw.startswith(b"\\xff\\xd8") else "webp" if raw[:4] == b"RIFF" and raw[8:12] == b"WEBP" else suffix\n file = Path(stem + "-" + label + "." + ext)\n file.write_bytes(raw)\n print("Saved", file)\n for index, block in enumerate(data.get("content") or []):\n print("Content", index, json.dumps(block, ensure_ascii=False))\n for choice in data.get("choices") or []:\n print("Choice", choice.get("index"), json.dumps(choice.get("message", choice.get("delta", {})), ensure_ascii=False))\n for index, item in enumerate(data.get("output") or []):\n print("Output", index, json.dumps(item, ensure_ascii=False))\n if item.get("type") == "image_generation_call" and item.get("result"):\n save_blob(item["result"], item.get("output_format") or "bin", "output-" + str(index))\n for ci, candidate in enumerate(data.get("candidates") or []):\n for pi, part in enumerate((candidate.get("content") or {}).get("parts") or []):\n print("Candidate/part", ci, pi, json.dumps(part, ensure_ascii=False))\n blob = part.get("inlineData") or part.get("inline_data")\n if blob and blob.get("data"):\n mime = blob.get("mimeType") or blob.get("mime_type")\n suffix = {"image/png":"png", "image/jpeg":"jpg", "image/webp":"webp", "audio/wav":"wav", "audio/mpeg":"mp3", "video/mp4":"mp4"}.get(mime, "bin")\n save_blob(blob["data"], suffix, "candidate-" + str(ci) + "-part-" + str(pi))\n\nsave_result(result.to_dict(mode="json"))\nfrom openai import OpenAI\nimport os\nfrom pathlib import Path\n\nif not os.environ.get("AILOOT_API_KEY", "").strip(): raise SystemExit("Set AILOOT_API_KEY before making a request")\nclient=OpenAI(base_url="https://api.ailoot.ai/v1",api_key=os.environ["AILOOT_API_KEY"],max_retries=0)\nimport base64, json\nfrom pathlib import Path\n\ndef save_result(data, stem="result"):\n # Keep every field, including usage, IDs, reasoning and tool requests.\n Path(stem + ".json").write_text(json.dumps(data, ensure_ascii=False, indent=2), encoding="utf-8")\n print(json.dumps(data, ensure_ascii=False, indent=2))\n def save_blob(encoded, suffix, label):\n raw = base64.b64decode(encoded, altchars=b"-_", validate=True)\n ext = "png" if raw.startswith(b"\\x89PNG\\r\\n\\x1a\\n") else "jpg" if raw.startswith(b"\\xff\\xd8") else "webp" if raw[:4] == b"RIFF" and raw[8:12] == b"WEBP" else suffix\n file = Path(stem + "-" + label + "." + ext)\n file.write_bytes(raw)\n print("Saved", file)\n for index, block in enumerate(data.get("content") or []):\n print("Content", index, json.dumps(block, ensure_ascii=False))\n for choice in data.get("choices") or []:\n print("Choice", choice.get("index"), json.dumps(choice.get("message", choice.get("delta", {})), ensure_ascii=False))\n for index, item in enumerate(data.get("output") or []):\n print("Output", index, json.dumps(item, ensure_ascii=False))\n if item.get("type") == "image_generation_call" and item.get("result"):\n save_blob(item["result"], item.get("output_format") or "bin", "output-" + str(index))\n for ci, candidate in enumerate(data.get("candidates") or []):\n for pi, part in enumerate((candidate.get("content") or {}).get("parts") or []):\n print("Candidate/part", ci, pi, json.dumps(part, ensure_ascii=False))\n blob = part.get("inlineData") or part.get("inline_data")\n if blob and blob.get("data"):\n mime = blob.get("mimeType") or blob.get("mime_type")\n suffix = {"image/png":"png", "image/jpeg":"jpg", "image/webp":"webp", "audio/wav":"wav", "audio/mpeg":"mp3", "video/mp4":"mp4"}.get(mime, "bin")\n save_blob(blob["data"], suffix, "candidate-" + str(ci) + "-part-" + str(pi))\n\nwith client.chat.completions.stream(model='qwen/qwen3.5-122b-a10b',messages=[{"role":"user","content":"Explain observations and hypotheses in three sentences."}]) as stream:\n with Path("stream-events.jsonl").open("w",encoding="utf-8") as log:\n for event in stream:\n payload=event.to_dict(mode="json")\n log.write(json.dumps(payload,ensure_ascii=False)+"\\n"); log.flush()\n print(json.dumps(payload,ensure_ascii=False))\n final=stream.get_final_completion()\n save_result(final.to_dict(mode="json"),"stream-result")\njson or SSE
Keep every complete returned response, file and content part.
Keep every choice and its content/tool calls/reasoning. In SSE assemble deltas by choice/tool index, retain the final finish reason and usage.
Keep request/response IDs and headers. No generic text job retrieval URL is supplied.
| Field | Type |
|---|---|
Fieldchoices[].index | Typeinteger |
Fieldchoices[].message.content | Typestring|null |
Fieldchoices[].message.tool_calls | Typetool_call[] |
Fieldchoices[].message.reasoning_content | Typestring |
Fieldchoices[].message.reasoning_details | Typeobject|array |
Fieldchoices[].finish_reason | Typestring|null |
Fieldusage | Typeobject |
| Response | Condition | Next step |
|---|---|---|
Response400 · invalid_request_error | ConditionInvalid required fields, material representation or parameter value | Next stepCorrect the indicated fields before a new creation request. |
Ailoot Ultra qwen/qwen3.5-122b-a10b
Use the method paths shown above.
qwen/qwen3.5-122b-a10bQwen3.5-122B-A10B is Qwen’s vision-language mixture-of-experts model with 122 billion total and 10 billion activated parameters. It takes text, images and video for reasoning, coding and agent tasks, returning text with selectable thinking and a native 262,144-token context.
This description covers the model’s capabilities. Available in Ailoot inputs, outputs and options are listed for the selected method below.
Qwen3.5 highlights: the 122B-A10B expert model
Qwen3.5-122B-A10B is a post-trained vision-language model for joint text and visual tasks. It accepts text, images and video and returns text, including explanations, code proposals and requested document artifacts. The family is built through early text–vision fusion and reinforcement learning on progressively complex tasks, developing reasoning, coding, agents and visual understanding. It describes coverage of 201 languages and dialects, with no implication that every language or domain has identical reliability.
A useful task can combine a technical specification, diagrams and relevant code, asking the model to explain the components and return an implementation proposal. A document-analysis task can ask for a comparison with supporting section or page references.
Model overview
The sparse mixture-of-experts checkpoint has 122 billion parameters in total and activates ten billion. It has 48 language-model layers, with groups of three Gated DeltaNet expert layers followed by one gated-attention expert layer. The maker lists 256 routed experts, selecting eight routed experts together with one shared expert. A vision encoder and multi-token prediction are part of the model. Its total and activated counts have different meanings; the activated figure is not the complete checkpoint size.
This is a distinct edition from dense Qwen3.5-27B, sparse Qwen3.5-35B-A3B and later model generations. Do not infer a general performance or latency ranking from the size alone. The native context is 262,144 tokens and the documented extension reaches 1,010,000 with configuration. Qwen advises retaining at least a 128K context where possible for complex thinking, while recognizing that memory pressure may require a smaller window.
Preparing joint text and visual evidence
Preserve diagram labels, chart axes, table headings and units. State whether the result should explain a relationship, extract named fields or propose code. When several sources matter, label them and request an output organized by the same identifiers. Separate visible facts from assumptions so a reviewer can trace the reasoning to the supplied material.
For a coding request, identify the intended behavior, relevant source and an input that demonstrates the issue. Ask for an editable patch with an explanation of the affected conditions and checks. For video, preserve the event sequence and timing and specify the question rather than request an undifferentiated summary. The model’s text result is a draft artifact to inspect; visual input capability does not mean image, video or speech generation.
Thinking and history
Thinking is enabled by default, with reasoning before the final answer. The maker’s enable_thinking setting can disable it for direct responses. The old /think and /nothink soft switches are not officially supported. Qwen3.5 recommends keeping final answers without thinking content in historical assistant messages. This should not be replaced by later Qwen generations’ historical preservation options simply because they share a family name.
Keep the final result easy to locate, such as a defined answer field, a delimited value or a separate code artifact. A direct answer still needs checking, while longer reasoning is useful only when its steps support a verifiable conclusion.
Agentic usage
Qwen documents tool calling and recommends Qwen-Agent with configured or MCP-connected tools, along with Qwen Code for coding-agent work. Define the permitted actions, available evidence and completion criteria. Connected tools execute operations and return results; the model uses that feedback in its analysis. A proposed command, retrieval query or patch should remain distinct from an action actually completed by the application.
Ultra-long context and video
The maker’s extension uses YaRN with the checkpoint’s interleaved positional settings. A factor of four targets the roughly one-million-token workflow, and two is suggested for 524,288-token workloads. Static scaling can affect shorter-text performance, so it should follow a real need for larger context. Input, represented visuals, history, reasoning and the final response all use the configured sequence budget.
The supported video example defaults to two frames per second. For hour-scale material, the maker recommends increasing the processor budget to roughly 224K video tokens for higher-rate sampling. That is a configuration-dependent recipe, not a universal media-duration, complete-recall or timestamp-precision guarantee.
Generation guidance and substantive limits
Best Practices recommend temperature 1.0, top_p 0.95, top_k 20, min_p 0, presence penalty 1.5 and repetition penalty 1.0 for general thinking. Precise coding uses temperature 0.6 and presence penalty zero. General direct responses use temperature 0.7, top_p 0.8 and presence penalty 1.5; top_k is 20, min_p is 0 and repetition penalty is 1.0. Direct-response reasoning has a separate preset: temperature 1.0, top_p 1.0, top_k 40 and presence penalty 2.0. Higher presence penalties can mix languages or slightly reduce quality.
Recommended generation allowances are 32,768 tokens for most queries and 81,920 for unusually complex mathematical or programming work; they are not separately established hard output ceilings. Define the requested format, retain source identifiers and check calculations, extracted values and proposed code against the original evidence. The useful outcome is a complete textual artifact whose assumptions and supporting material remain visible.
POST /v1/messagestextjson, sse| Field | Type | Requirement | Values & limits | Default |
|---|---|---|---|---|
Fieldmodel | Typestring | RequirementRequired | Values & limits qwen/qwen3.5-122b-a10b | Default— |
Fieldmessages | Typemessage[] | RequirementRequired | Values & limits | Default— |
Fieldmessages[].role | Typestring | RequirementRequired | Values & limits user, assistant | Default— |
Fieldmessages[].content | Typestring | RequirementRequired | Values & limits | Default— |
Fieldmax_tokens | Typeinteger | RequirementRequired | Values & limits | Default— |
Fieldstream | Typeboolean | RequirementOptional | Values & limits | Defaultfalse |
Text · UTF-8 text
messages[].content string
| Parameter | Contract |
|---|---|
Parametercontent_type | Contractapplication/json |
Parameterrequired | Contractmodel, messages, max_tokens |
Parameteroptional_controls | Contractstream |
Parametermaterial_types | Contracttext |
Parametertext_file_material | ContractRead the actual UTF-8 document and put its text in the indicated input or message content string. |
Lifecycle: synchronous_or_streamed
Tariff references do not add request options. Supported parameters are listed for the selected method.
Match each charge to the conditions shown. Use one rate for each set of alternative configurations.
| Configuration | Rate schedule | Context tokens, more than | Context tokens, up to and including | Charge |
|---|---|---|---|---|
| ConfigurationBase token rates | Rate scheduleBase rate | Context tokens, more than— | Context tokens, up to and includingNo upper bound | Charge Input · Input tokens $0.00058 / 1K tokens Output · Output tokens $0.00458 / 1K tokens Input · Cache read $0 / 1K tokens Input · Cache write $0 / 1K tokens Input · Cache write, 1 hour — / 1K tokens |
| ConfigurationRequest-length tier 1 | Rate schedule— | Context tokens, more than0 | Context tokens, up to and includingNo upper bound | Charge Input · Input tokens $0.00058 / 1K tokens Output · Output tokens $0.00458 / 1K tokens Input · Cache read $0 / 1K tokens Input · Cache write $0 / 1K tokens Input · Cache write, 1 hour — / 1K tokens |
Ailoot reserves an estimated amount before a request. After the result, it settles the final charge from measured usage. Streaming settlement uses the final usage record. How reservation and settlement work →
{\n "model": "qwen/qwen3.5-122b-a10b",\n "max_tokens": 128,\n "messages": [\n {\n "role": "user",\n "content": "Summarize the incident report."\n }\n ],\n "stream": false\n}set -eu\n: "${AILOOT_API_KEY:?Set your Ailoot API key}"\n: "${AILOOT_TEXT_FILE:?Set your actual UTF-8 document}"\nexport AILOOT_TEXT_FILE\npython3 - <<'PY_REQUEST'\nimport json, os\nfrom pathlib import Path\n\nsource_file = Path(os.environ['AILOOT_TEXT_FILE'])\nif not source_file.is_file():\n raise SystemExit('Set AILOOT_TEXT_FILE to your actual UTF-8 document')\ntext = source_file.read_text(encoding='utf-8')\nif not text.strip():\n raise SystemExit('Choose a document containing text')\npayload = {'model': 'qwen/qwen3.5-122b-a10b', 'max_tokens': 128, 'messages': [{'role': 'user', 'content': 'Summarize the incident report.'}]}\npayload['messages'][0]['content'] = 'Summarize this material and quote supporting source passages.\\n\\n' + text\nPath('request.json').write_text(json.dumps(payload, ensure_ascii=False), encoding='utf-8')\nPY_REQUEST\ncurl --fail-with-body --silent --show-error https://api.ailoot.ai/v1/messages \\\n -H "x-api-key: $AILOOT_API_KEY" -H 'anthropic-version: 2023-06-01' -H 'Content-Type: application/json' \\\n --data-binary @request.json -D response-headers.txt -o result.json\npython3 - <<'PY_RESULT'\nimport json\nfrom pathlib import Path\n\nresult = json.loads(Path('result.json').read_text(encoding='utf-8'))\nprint(json.dumps(result, ensure_ascii=False, indent=2))\nPY_RESULT\nimport json, os\nfrom pathlib import Path\n\nsource_file = Path(os.environ['AILOOT_TEXT_FILE'])\nif not source_file.is_file():\n raise SystemExit('Set AILOOT_TEXT_FILE to your actual UTF-8 document')\ntext = source_file.read_text(encoding='utf-8')\nif not text.strip():\n raise SystemExit('Choose a document containing text')\npayload = {'model': 'qwen/qwen3.5-122b-a10b', 'max_tokens': 128, 'messages': [{'role': 'user', 'content': 'Summarize the incident report.'}]}\npayload['messages'][0]['content'] = 'Summarize this material and quote supporting source passages.\\n\\n' + text\nfrom urllib.request import Request, urlopen\nfrom urllib.error import HTTPError\n\napi_key = os.environ['AILOOT_API_KEY']\nif not api_key.strip():\n raise SystemExit('Set AILOOT_API_KEY before making a request')\nheaders = {'x-api-key': 'PLACEHOLDER_KEY', 'anthropic-version': '2023-06-01', 'Content-Type': 'application/json'}\nheaders['x-api-key'] = api_key\nrequest = Request('https://api.ailoot.ai/v1/messages', data=json.dumps(payload, ensure_ascii=False).encode('utf-8'), headers=headers, method='POST')\ntry:\n with urlopen(request, timeout=120) as response:\n raw = response.read()\n Path('response-headers.txt').write_text(str(response.headers), encoding='utf-8')\nexcept HTTPError as error:\n Path('result.json').write_bytes(error.read())\n Path('response-headers.txt').write_text(str(error.headers), encoding='utf-8')\n raise\nPath('result.json').write_bytes(raw)\nimport json\nfrom pathlib import Path\n\nresult = json.loads(Path('result.json').read_text(encoding='utf-8'))\nprint(json.dumps(result, ensure_ascii=False, indent=2))\nJSON or SSE
Keep every complete returned response, file and content part.
Keep all content blocks, thinking signatures and actual tool-use IDs. Assemble all SSE block deltas; a first text block is not the complete output.
Keep request/response IDs and headers. No generic text job retrieval URL is supplied.
| Field | Type |
|---|---|
Fieldid | Typestring |
Fieldmodel | Typestring |
Fieldcontent[] | Typetext|thinking|tool_use|other content block |
Fieldstop_reason | Typestring|null |
Fieldusage | Typeobject |
| Response | Condition | Next step |
|---|---|---|
Response400 · invalid_request_error | ConditionInvalid required fields, material representation or parameter value | Next stepCorrect the indicated fields before a new creation request. |
Ailoot Ultra qwen/qwen3.5-122b-a10b
Use the method paths shown above.
Technical example in English
Check a process graph against its rule\n\nA process specification says validation must complete before persistence. Textual graph G1 is receive -> persist -> validate -> acknowledge. Code excerpt: save(record); validate(record); return "accepted". Propose an implementation correction and update the graph in text. Include a test showing that an invalid record never reaches save.A corrected sequence diagram in text, a focused code proposal and a test using an invalid record.Check that the graph and code agree on validate-before-save and that acknowledgment follows successful persistence.