Llama 3.1 Nemotron Safety Guard 8B v3 was developed by NVIDIA to moderate interaction between a person and a language model. A user message can be supplied alone or accompanied by an assistant response, including multiple turns. The model determines each side's safety: a dangerous request and an appropriate assistant refusal receive different assessments. This allows input to be checked before generation and a prepared response to be analyzed separately.
The result is JSON. The required User Safety field contains safe or unsafe. Response Safety appears when an assistant response is included in the input. If a violation is detected, Safety Categories is added with a comma-separated list of categories. This result is suited to programmatic filtering and referral of disputed cases for further review.
The instruction specifies the rules and category names used for the check. The published taxonomy has 23 risk groups, including violence, threats, harassment, disclosure of personal data, fraud and malicious code. Names in the result depend on the supplied policy and therefore need to be interpreted alongside the rules of the particular task.
Training accounts for regional differences in sensitive topics. Data was adapted to local expressions and cultural context, and inconsistent translations were discarded. Preparation included provocative requests and attempts to induce the assistant to break the rules. This training setup is particularly useful where a literal translation of prohibited words is insufficient to assess a message's meaning.
The nine main languages are English, Spanish, Chinese, German, French, Hindi, Japanese, Arabic and Thai. NVIDIA also describes generalization to other languages, including Russian; this is a separate property, not inclusion of Russian in the main nine. The common instruction is retained across languages, while the messages being checked are supplied in the conversation's language.
Input is text, with context of up to 8K tokens. The model is based on Llama 3.1 8B Instruct, fine-tuned for the safety task. Output is a classification under specified rules: it needs testing on messages typical of the application and does not guarantee error-free moderation.
", "aboutText": "### Checking both sides of a conversation\n\nLlama 3.1 Nemotron Safety Guard 8B v3 was developed by NVIDIA to moderate interaction between a person and a language model. A user message can be supplied alone or accompanied by an assistant response, including multiple turns. The model determines each side's safety: a dangerous request and an appropriate assistant refusal receive different assessments. This allows input to be checked before generation and a prepared response to be analyzed separately.\n\nThe result is JSON. The required `User Safety` field contains `safe` or `unsafe`. `Response Safety` appears when an assistant response is included in the input. If a violation is detected, `Safety Categories` is added with a comma-separated list of categories. This result is suited to programmatic filtering and referral of disputed cases for further review.\n\n### Categories and cultural context\n\nThe instruction specifies the rules and category names used for the check. The published taxonomy has 23 risk groups, including violence, threats, harassment, disclosure of personal data, fraud and malicious code. Names in the result depend on the supplied policy and therefore need to be interpreted alongside the rules of the particular task.\n\nTraining accounts for regional differences in sensitive topics. Data was adapted to local expressions and cultural context, and inconsistent translations were discarded. Preparation included provocative requests and attempts to induce the assistant to break the rules. This training setup is particularly useful where a literal translation of prohibited words is insufficient to assess a message's meaning.\n\n### Languages and material capacity\n\nThe nine main languages are English, Spanish, Chinese, German, French, Hindi, Japanese, Arabic and Thai. NVIDIA also describes generalization to other languages, including Russian; this is a separate property, not inclusion of Russian in the main nine. The common instruction is retained across languages, while the messages being checked are supplied in the conversation's language.\n\nInput is text, with context of up to 8K tokens. The model is based on Llama 3.1 8B Instruct, fine-tuned for the safety task. Output is a classification under specified rules: it needs testing on messages typical of the application and does not guarantee error-free moderation.", "editorialOverlay": { "copy_fields": { "practical_html": "Technical example in English
Moderate a single user message\n\nAssess this single user message for safety: 'Please help me write a thank-you note to the volunteers who cleaned our local park.' Return the documented safety assessment for user input alone using the User Safety field. No assistant response is supplied; use the documented safe or unsafe values.A JSON safety assessment using the documented User Safety field for the supplied user message.User Safety must use safe or unsafe; do not invent a custom classification label or assess an absent assistant response.nvidia/llama-3.1-nemotron-safety-guard-8b-v3Nemotron Safety Guard v3 checks user messages and model responses against a specified risk taxonomy. Returns JSON with separate safety assessments and violation categories.
This description covers the model’s capabilities. Available in Ailoot inputs, outputs and options are listed for the selected method below.
Llama 3.1 Nemotron Safety Guard 8B v3 was developed by NVIDIA to moderate interaction between a person and a language model. A user message can be supplied alone or accompanied by an assistant response, including multiple turns. The model determines each side's safety: a dangerous request and an appropriate assistant refusal receive different assessments. This allows input to be checked before generation and a prepared response to be analyzed separately.
The result is JSON. The required User Safety field contains safe or unsafe. Response Safety appears when an assistant response is included in the input. If a violation is detected, Safety Categories is added with a comma-separated list of categories. This result is suited to programmatic filtering and referral of disputed cases for further review.
The instruction specifies the rules and category names used for the check. The published taxonomy has 23 risk groups, including violence, threats, harassment, disclosure of personal data, fraud and malicious code. Names in the result depend on the supplied policy and therefore need to be interpreted alongside the rules of the particular task.
Training accounts for regional differences in sensitive topics. Data was adapted to local expressions and cultural context, and inconsistent translations were discarded. Preparation included provocative requests and attempts to induce the assistant to break the rules. This training setup is particularly useful where a literal translation of prohibited words is insufficient to assess a message's meaning.
The nine main languages are English, Spanish, Chinese, German, French, Hindi, Japanese, Arabic and Thai. NVIDIA also describes generalization to other languages, including Russian; this is a separate property, not inclusion of Russian in the main nine. The common instruction is retained across languages, while the messages being checked are supplied in the conversation's language.
Input is text, with context of up to 8K tokens. The model is based on Llama 3.1 8B Instruct, fine-tuned for the safety task. Output is a classification under specified rules: it needs testing on messages typical of the application and does not guarantee error-free moderation.
POST /v1/chat/completionstexttext| Field | Type | Requirement | Values & limits | Default |
|---|---|---|---|---|
Fieldmodel | Typestring | RequirementRequired | Values & limits nvidia/llama-3.1-nemotron-safety-guard-8b-v3 | 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 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— |
Text · UTF8 text
messages[].content
| Parameter | Contract |
|---|---|
Parametermaterial_types | Contracttext |
Parameterrequired | Contractmodel, messages |
Parameteroptional_controls | Contractnull |
Parametercontent_type | Contractapplication/json |
Parametertext_file_material | Contractactual UTF-8 content inserted as text; file IDs/URLs need separate exact support |
Lifecycle: exact retained published method
Method shapes and IDs are exact; supplier capability is separate from admission.
POST /v1/chat/completionsAiloot 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: "${AILOOT_API_KEY:?Set your Ailoot API key}"\n: "${AILOOT_TEXT_FILE:?Set your actual UTF-8 source-material file}"\nexport AILOOT_TEXT_FILE\npython3 - <<'PY_REQUEST'\nimport json,os\nfrom pathlib import Path\np=json.loads('{"model": "nvidia/llama-3.1-nemotron-safety-guard-8b-v3", "messages": []}')\nf=Path(os.environ["AILOOT_TEXT_FILE"])\nif not f.is_file():raise SystemExit("Choose your actual UTF-8 source material")\ntext=f.read_text(encoding="utf-8")\nif not text.strip():raise SystemExit("Source material must contain text")\np["messages"]=[{"role":"user","content":"Summarize this material and quote supporting source passages.\\n\\n"+text}]\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\nprint(json.dumps(json.loads(Path("result.json").read_text()),ensure_ascii=False,indent=2))\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)\nresp = client.chat.completions.create(\n model="nvidia/llama-3.1-nemotron-safety-guard-8b-v3",\n messages=[{"role": "user", "content": "Explain the difference between an observation and a hypothesis in three sentences."}],\n)\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(resp.to_dict(mode="json"))\nimport OpenAI from "openai";\n\nif (!process.env.AILOOT_API_KEY?.trim()) throw new Error("Set AILOOT_API_KEY before making a request");\nconst client = new OpenAI({ baseURL: "https://api.ailoot.ai/v1", apiKey: process.env.AILOOT_API_KEY, maxRetries: 0 });\nconst resp = await client.chat.completions.create({\n model: "nvidia/llama-3.1-nemotron-safety-guard-8b-v3",\n messages: [{ role: "user", content: "Explain the difference between an observation and a hypothesis in three sentences." }],\n});\nimport { writeFileSync } from "node:fs";\n\nfunction saveResult(data, stem = "result") {\n writeFileSync(stem + ".json", JSON.stringify(data, null, 2));\n console.log(JSON.stringify(data, null, 2));\n const saveBlob = (encoded, suffix, label) => {\n const raw = Buffer.from(encoded, "base64");\n const ext = raw.subarray(0,8).equals(Buffer.from([137,80,78,71,13,10,26,10])) ? "png" : raw[0]===255 && raw[1]===216 ? "jpg" : raw.subarray(0,4).toString()==="RIFF" && raw.subarray(8,12).toString()==="WEBP" ? "webp" : suffix;\n const file = `${stem}-${label}.${ext}`;\n writeFileSync(file, raw); console.log("Saved", file);\n };\n for (const [i, block] of (data.content || []).entries()) console.log("Content", i, JSON.stringify(block));\n for (const choice of data.choices || []) console.log("Choice", choice.index, JSON.stringify(choice.message || choice.delta || {}));\n for (const [i, item] of (data.output || []).entries()) {\n console.log("Output", i, JSON.stringify(item));\n if (item.type === "image_generation_call" && item.result) saveBlob(item.result, item.output_format || "bin", `output-${i}`);\n }\n for (const [ci, candidate] of (data.candidates || []).entries()) {\n for (const [pi, part] of (candidate.content?.parts || []).entries()) {\n console.log("Candidate/part", ci, pi, JSON.stringify(part));\n const blob = part.inlineData || part.inline_data;\n if (blob?.data) {\n const mime = blob.mimeType || blob.mime_type;\n const suffix = ({"image/png":"png", "image/jpeg":"jpg", "image/webp":"webp", "audio/wav":"wav", "audio/mpeg":"mp3", "video/mp4":"mp4"})[mime] || "bin";\n saveBlob(blob.data, suffix, `candidate-${ci}-part-${pi}`);\n }\n }\n }\n}\n\nsaveResult(resp);\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='nvidia/llama-3.1-nemotron-safety-guard-8b-v3',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 nvidia/llama-3.1-nemotron-safety-guard-8b-v3, nemotron-safety-guard, nemotron-safety-guard-latest
Use the method paths shown above.
nvidia/llama-3.1-nemotron-safety-guard-8b-v3, nemotron-safety-guard, nemotron-safety-guard-latest
nemotron-safety-guardNemotron Safety Guard v3 checks user messages and model responses against a specified risk taxonomy. Returns JSON with separate safety assessments and violation categories.
This description covers the model’s capabilities. Available in Ailoot inputs, outputs and options are listed for the selected method below.
Llama 3.1 Nemotron Safety Guard 8B v3 was developed by NVIDIA to moderate interaction between a person and a language model. A user message can be supplied alone or accompanied by an assistant response, including multiple turns. The model determines each side's safety: a dangerous request and an appropriate assistant refusal receive different assessments. This allows input to be checked before generation and a prepared response to be analyzed separately.
The result is JSON. The required User Safety field contains safe or unsafe. Response Safety appears when an assistant response is included in the input. If a violation is detected, Safety Categories is added with a comma-separated list of categories. This result is suited to programmatic filtering and referral of disputed cases for further review.
The instruction specifies the rules and category names used for the check. The published taxonomy has 23 risk groups, including violence, threats, harassment, disclosure of personal data, fraud and malicious code. Names in the result depend on the supplied policy and therefore need to be interpreted alongside the rules of the particular task.
Training accounts for regional differences in sensitive topics. Data was adapted to local expressions and cultural context, and inconsistent translations were discarded. Preparation included provocative requests and attempts to induce the assistant to break the rules. This training setup is particularly useful where a literal translation of prohibited words is insufficient to assess a message's meaning.
The nine main languages are English, Spanish, Chinese, German, French, Hindi, Japanese, Arabic and Thai. NVIDIA also describes generalization to other languages, including Russian; this is a separate property, not inclusion of Russian in the main nine. The common instruction is retained across languages, while the messages being checked are supplied in the conversation's language.
Input is text, with context of up to 8K tokens. The model is based on Llama 3.1 8B Instruct, fine-tuned for the safety task. Output is a classification under specified rules: it needs testing on messages typical of the application and does not guarantee error-free moderation.
POST /v1/chat/completionstexttext| Field | Type | Requirement | Values & limits | Default |
|---|---|---|---|---|
Fieldmodel | Typestring | RequirementRequired | Values & limits nemotron-safety-guard | 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 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— |
Text · UTF8 text
messages[].content
| Parameter | Contract |
|---|---|
Parametermaterial_types | Contracttext |
Parameterrequired | Contractmodel, messages |
Parameteroptional_controls | Contractnull |
Parametercontent_type | Contractapplication/json |
Parametertext_file_material | Contractactual UTF-8 content inserted as text; file IDs/URLs need separate exact support |
Lifecycle: exact retained published method
Method shapes and IDs are exact; supplier capability is separate from admission.
POST /v1/chat/completionsAiloot 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}"\n: "${AILOOT_TEXT_FILE:?Set your actual UTF-8 source-material file}"\nexport AILOOT_TEXT_FILE\npython3 - <<'PY_REQUEST'\nimport json,os\nfrom pathlib import Path\np=json.loads('{"model": "nemotron-safety-guard", "messages": []}')\nf=Path(os.environ["AILOOT_TEXT_FILE"])\nif not f.is_file():raise SystemExit("Choose your actual UTF-8 source material")\ntext=f.read_text(encoding="utf-8")\nif not text.strip():raise SystemExit("Source material must contain text")\np["messages"]=[{"role":"user","content":"Summarize this material and quote supporting source passages.\\n\\n"+text}]\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\nprint(json.dumps(json.loads(Path("result.json").read_text()),ensure_ascii=False,indent=2))\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)\nresp = client.chat.completions.create(\n model="nemotron-safety-guard",\n messages=[{"role": "user", "content": "Explain the difference between an observation and a hypothesis in three sentences."}],\n)\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(resp.to_dict(mode="json"))\nimport OpenAI from "openai";\n\nif (!process.env.AILOOT_API_KEY?.trim()) throw new Error("Set AILOOT_API_KEY before making a request");\nconst client = new OpenAI({ baseURL: "https://api.ailoot.ai/v1", apiKey: process.env.AILOOT_API_KEY, maxRetries: 0 });\nconst resp = await client.chat.completions.create({\n model: "nemotron-safety-guard",\n messages: [{ role: "user", content: "Explain the difference between an observation and a hypothesis in three sentences." }],\n});\nimport { writeFileSync } from "node:fs";\n\nfunction saveResult(data, stem = "result") {\n writeFileSync(stem + ".json", JSON.stringify(data, null, 2));\n console.log(JSON.stringify(data, null, 2));\n const saveBlob = (encoded, suffix, label) => {\n const raw = Buffer.from(encoded, "base64");\n const ext = raw.subarray(0,8).equals(Buffer.from([137,80,78,71,13,10,26,10])) ? "png" : raw[0]===255 && raw[1]===216 ? "jpg" : raw.subarray(0,4).toString()==="RIFF" && raw.subarray(8,12).toString()==="WEBP" ? "webp" : suffix;\n const file = `${stem}-${label}.${ext}`;\n writeFileSync(file, raw); console.log("Saved", file);\n };\n for (const [i, block] of (data.content || []).entries()) console.log("Content", i, JSON.stringify(block));\n for (const choice of data.choices || []) console.log("Choice", choice.index, JSON.stringify(choice.message || choice.delta || {}));\n for (const [i, item] of (data.output || []).entries()) {\n console.log("Output", i, JSON.stringify(item));\n if (item.type === "image_generation_call" && item.result) saveBlob(item.result, item.output_format || "bin", `output-${i}`);\n }\n for (const [ci, candidate] of (data.candidates || []).entries()) {\n for (const [pi, part] of (candidate.content?.parts || []).entries()) {\n console.log("Candidate/part", ci, pi, JSON.stringify(part));\n const blob = part.inlineData || part.inline_data;\n if (blob?.data) {\n const mime = blob.mimeType || blob.mime_type;\n const suffix = ({"image/png":"png", "image/jpeg":"jpg", "image/webp":"webp", "audio/wav":"wav", "audio/mpeg":"mp3", "video/mp4":"mp4"})[mime] || "bin";\n saveBlob(blob.data, suffix, `candidate-${ci}-part-${pi}`);\n }\n }\n }\n}\n\nsaveResult(resp);\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='nemotron-safety-guard',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 nvidia/llama-3.1-nemotron-safety-guard-8b-v3, nemotron-safety-guard, nemotron-safety-guard-latest
Use the method paths shown above.
nvidia/llama-3.1-nemotron-safety-guard-8b-v3, nemotron-safety-guard, nemotron-safety-guard-latest
nemotron-safety-guard-latestNemotron Safety Guard v3 checks user messages and model responses against a specified risk taxonomy. Returns JSON with separate safety assessments and violation categories.
This description covers the model’s capabilities. Available in Ailoot inputs, outputs and options are listed for the selected method below.
Llama 3.1 Nemotron Safety Guard 8B v3 was developed by NVIDIA to moderate interaction between a person and a language model. A user message can be supplied alone or accompanied by an assistant response, including multiple turns. The model determines each side's safety: a dangerous request and an appropriate assistant refusal receive different assessments. This allows input to be checked before generation and a prepared response to be analyzed separately.
The result is JSON. The required User Safety field contains safe or unsafe. Response Safety appears when an assistant response is included in the input. If a violation is detected, Safety Categories is added with a comma-separated list of categories. This result is suited to programmatic filtering and referral of disputed cases for further review.
The instruction specifies the rules and category names used for the check. The published taxonomy has 23 risk groups, including violence, threats, harassment, disclosure of personal data, fraud and malicious code. Names in the result depend on the supplied policy and therefore need to be interpreted alongside the rules of the particular task.
Training accounts for regional differences in sensitive topics. Data was adapted to local expressions and cultural context, and inconsistent translations were discarded. Preparation included provocative requests and attempts to induce the assistant to break the rules. This training setup is particularly useful where a literal translation of prohibited words is insufficient to assess a message's meaning.
The nine main languages are English, Spanish, Chinese, German, French, Hindi, Japanese, Arabic and Thai. NVIDIA also describes generalization to other languages, including Russian; this is a separate property, not inclusion of Russian in the main nine. The common instruction is retained across languages, while the messages being checked are supplied in the conversation's language.
Input is text, with context of up to 8K tokens. The model is based on Llama 3.1 8B Instruct, fine-tuned for the safety task. Output is a classification under specified rules: it needs testing on messages typical of the application and does not guarantee error-free moderation.
POST /v1/chat/completionstexttext| Field | Type | Requirement | Values & limits | Default |
|---|---|---|---|---|
Fieldmodel | Typestring | RequirementRequired | Values & limits nemotron-safety-guard-latest | 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 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— |
Text · UTF8 text
messages[].content
| Parameter | Contract |
|---|---|
Parametermaterial_types | Contracttext |
Parameterrequired | Contractmodel, messages |
Parameteroptional_controls | Contractnull |
Parametercontent_type | Contractapplication/json |
Parametertext_file_material | Contractactual UTF-8 content inserted as text; file IDs/URLs need separate exact support |
Lifecycle: exact retained published method
Method shapes and IDs are exact; supplier capability is separate from admission.
POST /v1/chat/completionsAiloot 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}"\n: "${AILOOT_TEXT_FILE:?Set your actual UTF-8 source-material file}"\nexport AILOOT_TEXT_FILE\npython3 - <<'PY_REQUEST'\nimport json,os\nfrom pathlib import Path\np=json.loads('{"model": "nemotron-safety-guard-latest", "messages": []}')\nf=Path(os.environ["AILOOT_TEXT_FILE"])\nif not f.is_file():raise SystemExit("Choose your actual UTF-8 source material")\ntext=f.read_text(encoding="utf-8")\nif not text.strip():raise SystemExit("Source material must contain text")\np["messages"]=[{"role":"user","content":"Summarize this material and quote supporting source passages.\\n\\n"+text}]\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\nprint(json.dumps(json.loads(Path("result.json").read_text()),ensure_ascii=False,indent=2))\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)\nresp = client.chat.completions.create(\n model="nemotron-safety-guard-latest",\n messages=[{"role": "user", "content": "Explain the difference between an observation and a hypothesis in three sentences."}],\n)\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(resp.to_dict(mode="json"))\nimport OpenAI from "openai";\n\nif (!process.env.AILOOT_API_KEY?.trim()) throw new Error("Set AILOOT_API_KEY before making a request");\nconst client = new OpenAI({ baseURL: "https://api.ailoot.ai/v1", apiKey: process.env.AILOOT_API_KEY, maxRetries: 0 });\nconst resp = await client.chat.completions.create({\n model: "nemotron-safety-guard-latest",\n messages: [{ role: "user", content: "Explain the difference between an observation and a hypothesis in three sentences." }],\n});\nimport { writeFileSync } from "node:fs";\n\nfunction saveResult(data, stem = "result") {\n writeFileSync(stem + ".json", JSON.stringify(data, null, 2));\n console.log(JSON.stringify(data, null, 2));\n const saveBlob = (encoded, suffix, label) => {\n const raw = Buffer.from(encoded, "base64");\n const ext = raw.subarray(0,8).equals(Buffer.from([137,80,78,71,13,10,26,10])) ? "png" : raw[0]===255 && raw[1]===216 ? "jpg" : raw.subarray(0,4).toString()==="RIFF" && raw.subarray(8,12).toString()==="WEBP" ? "webp" : suffix;\n const file = `${stem}-${label}.${ext}`;\n writeFileSync(file, raw); console.log("Saved", file);\n };\n for (const [i, block] of (data.content || []).entries()) console.log("Content", i, JSON.stringify(block));\n for (const choice of data.choices || []) console.log("Choice", choice.index, JSON.stringify(choice.message || choice.delta || {}));\n for (const [i, item] of (data.output || []).entries()) {\n console.log("Output", i, JSON.stringify(item));\n if (item.type === "image_generation_call" && item.result) saveBlob(item.result, item.output_format || "bin", `output-${i}`);\n }\n for (const [ci, candidate] of (data.candidates || []).entries()) {\n for (const [pi, part] of (candidate.content?.parts || []).entries()) {\n console.log("Candidate/part", ci, pi, JSON.stringify(part));\n const blob = part.inlineData || part.inline_data;\n if (blob?.data) {\n const mime = blob.mimeType || blob.mime_type;\n const suffix = ({"image/png":"png", "image/jpeg":"jpg", "image/webp":"webp", "audio/wav":"wav", "audio/mpeg":"mp3", "video/mp4":"mp4"})[mime] || "bin";\n saveBlob(blob.data, suffix, `candidate-${ci}-part-${pi}`);\n }\n }\n }\n}\n\nsaveResult(resp);\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='nemotron-safety-guard-latest',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 nvidia/llama-3.1-nemotron-safety-guard-8b-v3, nemotron-safety-guard, nemotron-safety-guard-latest
Use the method paths shown above.
nvidia/llama-3.1-nemotron-safety-guard-8b-v3, nemotron-safety-guard, nemotron-safety-guard-latest
Technical example in English
Moderate a single user message\n\nAssess this single user message for safety: 'Please help me write a thank-you note to the volunteers who cleaned our local park.' Return the documented safety assessment for user input alone using the User Safety field. No assistant response is supplied; use the documented safe or unsafe values.A JSON safety assessment using the documented User Safety field for the supplied user message.User Safety must use safe or unsafe; do not invent a custom classification label or assess an absent assistant response.