NVIDIA Nemotron 3 Ultra is the large-scale reasoning and orchestration model in the Nemotron 3 family. It is designed for demanding multi-step agents, long-context analysis and reasoning over code, mathematics and science. The model has 550B total and 55B active parameters. It combines Mamba-2, attention and Latent Mixture-of-Experts layers, with Multi-Token Prediction layers supporting its generation design.
LatentMoE projects information into a smaller latent representation for expert processing. NVIDIA’s training recipe combines pre-training, supervised fine-tuning, reinforcement learning and multi-domain on-policy distillation. The practical purpose of these stages is to improve reasoning, instruction following, tool use and agent behavior across several task types. Ultra produces text: an explanation, a solution, code, a plan or a requested response structure. Supply the material the answer should use and the conditions that determine whether the result is correct.
For mathematical or scientific reasoning, include the complete statement, known values, units and permitted assumptions. A useful request is: “Derive the result from these conditions, identify the assumptions at each stage, check the dimensions, and explain which observation would invalidate the conclusion.” Ask for a concise final result after the working explanation. If experimental data is involved, distinguish what was measured from what was calculated, and ask for uncertainty or missing information to remain visible rather than being absorbed into a confident answer.
For code and agent workflows, provide the objective, relevant files as text and the real outputs of prior steps. Ultra is intended for tasks involving several dependent decisions, so an explicit record of completed work and remaining questions is useful. For example: “Review this migration proposal against the schema and failing checks. Identify the dependency that blocks completion, propose a focused correction, and give the expected result of the next check.” A tool-connected application can then execute a proposed step and return evidence for the next decision.
The manufacturer supports reasoning on or off through the chat template. Reasoning is on by default, producing a reasoning trace followed by the answer. The Ultra card also documents a medium-effort reasoning mode that uses fewer reasoning tokens than full thinking. This is an Ultra-specific option; it should not be confused with the low-effort setting documented for Super. Set the desired final deliverable and leave space for it after the reasoning. An unfinished trace does not substitute for a completed answer to the original question.
Ultra supports up to a 1M-token context. For a large collection of documents or source files, label sections and tell the model exactly what to compare, trace or reconcile. Keep the source order when chronology or dependency matters. A useful long-document result contains the finding, the supporting passage and the implication for the question. Select the relevant materials and reserve output space; the context window describes the working sequence rather than a separate guaranteed final-answer length.
The maker lists English, French, Spanish, Italian, German, Japanese, Hindi, Korean, Brazilian Portuguese and Chinese. State the output language and preserve technical identifiers or source quotations where exact wording matters. For a multilingual analysis, use the same criteria for each document and ask for differences in terminology or assumptions to be explained. Fluency alone is not enough to establish that two differently worded requirements mean the same thing.
NVIDIA provides explicit tool-calling examples with reasoning. Define each function’s purpose, arguments and expected result, then return its actual output to the model. The application performs the operation; the model reasons over the returned observation. A long agent workflow becomes easier to inspect when proposed actions, executed results and unresolved checks are clearly distinguished. For a final report, request the answer, the evidence that supports it, the assumptions and any remaining task items. Verify calculations against the original units, code changes against observed checks, and factual claims against the supplied sources.
", "aboutText": "NVIDIA Nemotron 3 Ultra is the large-scale reasoning and orchestration model in the Nemotron 3 family. It is designed for demanding multi-step agents, long-context analysis and reasoning over code, mathematics and science. The model has 550B total and 55B active parameters. It combines Mamba-2, attention and Latent Mixture-of-Experts layers, with Multi-Token Prediction layers supporting its generation design.\n\nLatentMoE projects information into a smaller latent representation for expert processing. NVIDIA’s training recipe combines pre-training, supervised fine-tuning, reinforcement learning and multi-domain on-policy distillation. The practical purpose of these stages is to improve reasoning, instruction following, tool use and agent behavior across several task types. Ultra produces text: an explanation, a solution, code, a plan or a requested response structure. Supply the material the answer should use and the conditions that determine whether the result is correct.\n\nFor mathematical or scientific reasoning, include the complete statement, known values, units and permitted assumptions. A useful request is: “Derive the result from these conditions, identify the assumptions at each stage, check the dimensions, and explain which observation would invalidate the conclusion.” Ask for a concise final result after the working explanation. If experimental data is involved, distinguish what was measured from what was calculated, and ask for uncertainty or missing information to remain visible rather than being absorbed into a confident answer.\n\nFor code and agent workflows, provide the objective, relevant files as text and the real outputs of prior steps. Ultra is intended for tasks involving several dependent decisions, so an explicit record of completed work and remaining questions is useful. For example: “Review this migration proposal against the schema and failing checks. Identify the dependency that blocks completion, propose a focused correction, and give the expected result of the next check.” A tool-connected application can then execute a proposed step and return evidence for the next decision.\n\nThe manufacturer supports reasoning on or off through the chat template. Reasoning is on by default, producing a reasoning trace followed by the answer. The Ultra card also documents a medium-effort reasoning mode that uses fewer reasoning tokens than full thinking. This is an Ultra-specific option; it should not be confused with the low-effort setting documented for Super. Set the desired final deliverable and leave space for it after the reasoning. An unfinished trace does not substitute for a completed answer to the original question.\n\nUltra supports up to a 1M-token context. For a large collection of documents or source files, label sections and tell the model exactly what to compare, trace or reconcile. Keep the source order when chronology or dependency matters. A useful long-document result contains the finding, the supporting passage and the implication for the question. Select the relevant materials and reserve output space; the context window describes the working sequence rather than a separate guaranteed final-answer length.\n\nThe maker lists English, French, Spanish, Italian, German, Japanese, Hindi, Korean, Brazilian Portuguese and Chinese. State the output language and preserve technical identifiers or source quotations where exact wording matters. For a multilingual analysis, use the same criteria for each document and ask for differences in terminology or assumptions to be explained. Fluency alone is not enough to establish that two differently worded requirements mean the same thing.\n\nNVIDIA provides explicit tool-calling examples with reasoning. Define each function’s purpose, arguments and expected result, then return its actual output to the model. The application performs the operation; the model reasons over the returned observation. A long agent workflow becomes easier to inspect when proposed actions, executed results and unresolved checks are clearly distinguished. For a final report, request the answer, the evidence that supports it, the assumptions and any remaining task items. Verify calculations against the original units, code changes against observed checks, and factual claims against the supplied sources.", "editorialOverlay": { "copy_fields": null, "copy_sha256": null, "copy_source": null, "html_sha256": "170e0b1fe669d5d17051a248febdf9895c5541c5f72237e4e3bfa3345ad20fd1", "module_sha256": "ee9dc1c50b1e6a8fb7fa062030fcb4646b9aaaeb08be513b723a5a92a7290f06", "raw_model_unchanged": true, "schema": "ailoot-native-model-editorial-v1" }, "locale": "en", "methods": [ { "html": "nvidia/nemotron-3-ultra-550b-a55b:freeNVIDIA Nemotron 3 Ultra is a large text reasoning model with 550B total and 55B active parameters. Its hybrid LatentMoE design supports demanding multi-step agents, code, mathematics, science and long-document analysis, with up to a 1M-token context and maker-documented reasoning controls.
This description covers the model’s capabilities. Available in Ailoot inputs, outputs and options are listed for the selected method below.
NVIDIA Nemotron 3 Ultra is the large-scale reasoning and orchestration model in the Nemotron 3 family. It is designed for demanding multi-step agents, long-context analysis and reasoning over code, mathematics and science. The model has 550B total and 55B active parameters. It combines Mamba-2, attention and Latent Mixture-of-Experts layers, with Multi-Token Prediction layers supporting its generation design.
LatentMoE projects information into a smaller latent representation for expert processing. NVIDIA’s training recipe combines pre-training, supervised fine-tuning, reinforcement learning and multi-domain on-policy distillation. The practical purpose of these stages is to improve reasoning, instruction following, tool use and agent behavior across several task types. Ultra produces text: an explanation, a solution, code, a plan or a requested response structure. Supply the material the answer should use and the conditions that determine whether the result is correct.
For mathematical or scientific reasoning, include the complete statement, known values, units and permitted assumptions. A useful request is: “Derive the result from these conditions, identify the assumptions at each stage, check the dimensions, and explain which observation would invalidate the conclusion.” Ask for a concise final result after the working explanation. If experimental data is involved, distinguish what was measured from what was calculated, and ask for uncertainty or missing information to remain visible rather than being absorbed into a confident answer.
For code and agent workflows, provide the objective, relevant files as text and the real outputs of prior steps. Ultra is intended for tasks involving several dependent decisions, so an explicit record of completed work and remaining questions is useful. For example: “Review this migration proposal against the schema and failing checks. Identify the dependency that blocks completion, propose a focused correction, and give the expected result of the next check.” A tool-connected application can then execute a proposed step and return evidence for the next decision.
The manufacturer supports reasoning on or off through the chat template. Reasoning is on by default, producing a reasoning trace followed by the answer. The Ultra card also documents a medium-effort reasoning mode that uses fewer reasoning tokens than full thinking. This is an Ultra-specific option; it should not be confused with the low-effort setting documented for Super. Set the desired final deliverable and leave space for it after the reasoning. An unfinished trace does not substitute for a completed answer to the original question.
Ultra supports up to a 1M-token context. For a large collection of documents or source files, label sections and tell the model exactly what to compare, trace or reconcile. Keep the source order when chronology or dependency matters. A useful long-document result contains the finding, the supporting passage and the implication for the question. Select the relevant materials and reserve output space; the context window describes the working sequence rather than a separate guaranteed final-answer length.
The maker lists English, French, Spanish, Italian, German, Japanese, Hindi, Korean, Brazilian Portuguese and Chinese. State the output language and preserve technical identifiers or source quotations where exact wording matters. For a multilingual analysis, use the same criteria for each document and ask for differences in terminology or assumptions to be explained. Fluency alone is not enough to establish that two differently worded requirements mean the same thing.
NVIDIA provides explicit tool-calling examples with reasoning. Define each function’s purpose, arguments and expected result, then return its actual output to the model. The application performs the operation; the model reasons over the returned observation. A long agent workflow becomes easier to inspect when proposed actions, executed results and unresolved checks are clearly distinguished. For a final report, request the answer, the evidence that supports it, the assumptions and any remaining task items. Verify calculations against the original units, code changes against observed checks, and factual claims against the supplied sources.
POST /v1/chat/completionstexttext| Field | Type | Requirement | Values & limits | Default |
|---|---|---|---|---|
Fieldmodel | Typestring | RequirementRequired | Values & limits nvidia/nemotron-3-ultra-550b-a55b:free | Default— |
Fieldmessages | Typearray | RequirementRequired | Values & limits | Default— |
Fieldstream | Typeboolean | RequirementOptional | Values & limits false | Defaultfalse |
Fieldmax_tokens | Typeinteger | RequirementOptional | Values & limits Minimum: 1 Maximum: 65,536 | Default— |
Fieldmax_completion_tokens | Typeinteger | RequirementOptional | Values & limits Minimum: 1 Maximum: 65,536 | Default— |
Fieldn | Typeinteger | RequirementOptional | Values & limits Minimum: 1 Maximum: 128 | Default— |
Fieldtemperature | Typenumber | RequirementOptional | Values & limits | Default— |
Fieldtop_p | Typenumber | RequirementOptional | Values & limits | Default— |
Fieldseed | Typeinteger | RequirementOptional | Values & limits | Default— |
Fieldstop | Type | RequirementOptional | Values & limits | Default— |
Fieldfrequency_penalty | Typenumber | RequirementOptional | Values & limits | Default— |
Fieldpresence_penalty | Typenumber | RequirementOptional | Values & limits | Default— |
Fielduser | Typestring | RequirementOptional | Values & limits | Default— |
| Parameter | Contract |
|---|---|
Parametercontent_type | Contractapplication/json |
Parameterrequired | Contractmodel, messages |
Parameteroptional | Contractstream, max_tokens, max_completion_tokens, n, temperature, top_p, seed, stop, frequency_penalty, presence_penalty, user |
Parametermaterial_types | Contracttext |
Parameterconditions | ContractUse literal text messages only. Cache controls, tools, image/audio/video content and streamed responses are unavailable for this method., The complete encoded JSON request must not exceed 1000000 bytes. |
Lifecycle: synchronous_json
Tariff references do not add request options. Supported parameters are listed for the selected method.
| Configuration | Rate schedule | Charge |
|---|---|---|
| ConfigurationBase token rates | Rate scheduleBase rate | Charge Input · Input tokens $0 / 1K tokens Output · Output tokens $0 / 1K tokens Input · Cache read — / 1K tokens Input · Cache write — / 1K tokens Input · Cache write, 1 hour — / 1K tokens |
Ailoot reserves an estimated amount before a request. After the result, it settles the final charge from measured usage. Streaming settlement uses the final usage record. How reservation and settlement work →
set -eu\n: "${AILOOT_API_KEY:?Set your Ailoot API key}"\nexport AILOOT_API_KEY\npython3 - <<'PY_REQUEST'\nimport json,os\nfrom pathlib import Path\np=json.loads('{"model": "nvidia/nemotron-3-ultra-550b-a55b:free", "messages": [{"role": "user", "content": "Summarize the supplied text in three concise points."}], "stream": false}')\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\n# Submit once. Retain the complete result, original identifiers and every response header.\nJSON
Keep every complete returned response, file and content part.
Retain the complete response, all actual results, original identifiers, usage and response headers.
Preserve the original response, headers and X-Client-Request-ID. An uncertain outcome does not authorize another creation.
Ailoot Ultra nvidia/nemotron-3-ultra-550b-a55b:free
Use the method paths shown above.
nvidia/nemotron-3-ultra-550b-a55b:freeNVIDIA Nemotron 3 Ultra is a large text reasoning model with 550B total and 55B active parameters. Its hybrid LatentMoE design supports demanding multi-step agents, code, mathematics, science and long-document analysis, with up to a 1M-token context and maker-documented reasoning controls.
This description covers the model’s capabilities. Available in Ailoot inputs, outputs and options are listed for the selected method below.
NVIDIA Nemotron 3 Ultra is the large-scale reasoning and orchestration model in the Nemotron 3 family. It is designed for demanding multi-step agents, long-context analysis and reasoning over code, mathematics and science. The model has 550B total and 55B active parameters. It combines Mamba-2, attention and Latent Mixture-of-Experts layers, with Multi-Token Prediction layers supporting its generation design.
LatentMoE projects information into a smaller latent representation for expert processing. NVIDIA’s training recipe combines pre-training, supervised fine-tuning, reinforcement learning and multi-domain on-policy distillation. The practical purpose of these stages is to improve reasoning, instruction following, tool use and agent behavior across several task types. Ultra produces text: an explanation, a solution, code, a plan or a requested response structure. Supply the material the answer should use and the conditions that determine whether the result is correct.
For mathematical or scientific reasoning, include the complete statement, known values, units and permitted assumptions. A useful request is: “Derive the result from these conditions, identify the assumptions at each stage, check the dimensions, and explain which observation would invalidate the conclusion.” Ask for a concise final result after the working explanation. If experimental data is involved, distinguish what was measured from what was calculated, and ask for uncertainty or missing information to remain visible rather than being absorbed into a confident answer.
For code and agent workflows, provide the objective, relevant files as text and the real outputs of prior steps. Ultra is intended for tasks involving several dependent decisions, so an explicit record of completed work and remaining questions is useful. For example: “Review this migration proposal against the schema and failing checks. Identify the dependency that blocks completion, propose a focused correction, and give the expected result of the next check.” A tool-connected application can then execute a proposed step and return evidence for the next decision.
The manufacturer supports reasoning on or off through the chat template. Reasoning is on by default, producing a reasoning trace followed by the answer. The Ultra card also documents a medium-effort reasoning mode that uses fewer reasoning tokens than full thinking. This is an Ultra-specific option; it should not be confused with the low-effort setting documented for Super. Set the desired final deliverable and leave space for it after the reasoning. An unfinished trace does not substitute for a completed answer to the original question.
Ultra supports up to a 1M-token context. For a large collection of documents or source files, label sections and tell the model exactly what to compare, trace or reconcile. Keep the source order when chronology or dependency matters. A useful long-document result contains the finding, the supporting passage and the implication for the question. Select the relevant materials and reserve output space; the context window describes the working sequence rather than a separate guaranteed final-answer length.
The maker lists English, French, Spanish, Italian, German, Japanese, Hindi, Korean, Brazilian Portuguese and Chinese. State the output language and preserve technical identifiers or source quotations where exact wording matters. For a multilingual analysis, use the same criteria for each document and ask for differences in terminology or assumptions to be explained. Fluency alone is not enough to establish that two differently worded requirements mean the same thing.
NVIDIA provides explicit tool-calling examples with reasoning. Define each function’s purpose, arguments and expected result, then return its actual output to the model. The application performs the operation; the model reasons over the returned observation. A long agent workflow becomes easier to inspect when proposed actions, executed results and unresolved checks are clearly distinguished. For a final report, request the answer, the evidence that supports it, the assumptions and any remaining task items. Verify calculations against the original units, code changes against observed checks, and factual claims against the supplied sources.
POST /v1/messagestexttext| Field | Type | Requirement | Values & limits | Default |
|---|---|---|---|---|
Fieldmodel | Typestring | RequirementRequired | Values & limits nvidia/nemotron-3-ultra-550b-a55b:free | Default— |
Fieldmessages | Typearray | RequirementRequired | Values & limits | Default— |
Fieldstream | Typeboolean | RequirementOptional | Values & limits false | Defaultfalse |
Fieldsystem | Typestring | RequirementOptional | Values & limits | Default— |
Fieldmax_tokens | Typeinteger | RequirementRequired | Values & limits Minimum: 1 Maximum: 65,536 | Default— |
Fieldtemperature | Typenumber | RequirementOptional | Values & limits | Default— |
Fieldtop_p | Typenumber | RequirementOptional | Values & limits | Default— |
Fieldtop_k | Typeinteger | RequirementOptional | Values & limits | Default— |
Fieldstop_sequences | Typearray | RequirementOptional | Values & limits | Default— |
Fieldmetadata | Typeobject | RequirementOptional | Values & limits | Default— |
| Parameter | Contract |
|---|---|
Parametercontent_type | Contractapplication/json |
Parameterrequired | Contractmodel, messages, max_tokens |
Parameteroptional | Contractstream, system, temperature, top_p, top_k, stop_sequences, metadata |
Parametermaterial_types | Contracttext |
Parameterconditions | ContractUse literal text messages only. Cache controls, tools, image/audio/video content and streamed responses are unavailable for this method., The complete encoded JSON request must not exceed 1000000 bytes. |
Lifecycle: synchronous_json
Tariff references do not add request options. Supported parameters are listed for the selected method.
| Configuration | Rate schedule | Charge |
|---|---|---|
| ConfigurationBase token rates | Rate scheduleBase rate | Charge Input · Input tokens $0 / 1K tokens Output · Output tokens $0 / 1K tokens Input · Cache read — / 1K tokens Input · Cache write — / 1K tokens Input · Cache write, 1 hour — / 1K tokens |
Ailoot reserves an estimated amount before a request. After the result, it settles the final charge from measured usage. Streaming settlement uses the final usage record. How reservation and settlement work →
set -eu\n: "${AILOOT_API_KEY:?Set your Ailoot API key}"\nexport AILOOT_API_KEY\npython3 - <<'PY_REQUEST'\nimport json,os\nfrom pathlib import Path\np=json.loads('{"model": "nvidia/nemotron-3-ultra-550b-a55b:free", "messages": [{"role": "user", "content": "Summarize the supplied text in three concise points."}], "stream": false, "max_tokens": 256}')\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/messages \\\n -H "Authorization: Bearer $AILOOT_API_KEY" -H 'Content-Type: application/json' \\\n --data-binary @request.json -D response-headers.txt -o result.json\n# Submit once. Retain the complete result, original identifiers and every response header.\nJSON
Keep every complete returned response, file and content part.
Retain the complete response, all actual results, original identifiers, usage and response headers.
Preserve the original response, headers and X-Client-Request-ID. An uncertain outcome does not authorize another creation.
Ailoot Ultra nvidia/nemotron-3-ultra-550b-a55b:free
Use the method paths shown above.