Skip to main content
The Agent API supports two execution modes — Claude Agent (default) and Standard Chat — each with their own set of available models.

​
Agent Runtimes — Claude vs Codex

Within the Claude Agent mode (type: "claude"), the API runs your task on one of two agent runtimes, selected automatically from the model id:
RuntimeSDKWire protocolUsed for
Claude Agent (default)@anthropic-ai/claude-agent-sdkAnthropic Messages (/v1/messages)Anthropic/Claude ids and all other providers
Codex Agent@openai/codex-sdkOpenAI Responses (/v1/responses)OpenAI/Codex ids
Selection rule — a model id runs on the Codex runtime when it contains codex, openai/, gpt-, or starts with o1/o3/o4; otherwise it runs on the Claude runtime.
Example model idRuntime
blackboxai/openai/gpt-5.3-codexCodex
blackboxai/openai/gpt-5.5 (and gpt-5.4, gpt-5.4-pro, gpt-5.4-nano)Codex
blackboxai/anthropic/claude-opus-4.6 / claude-sonnet-4.5Claude
everything else (nemotron, minimax, glm, …)Claude
Both runtimes are fully managed — they run inside the same sandbox, expose the same task lifecycle and event stream, and are interchangeable from the caller’s perspective. You normally don’t pick the runtime directly; you pick the model and the runtime follows.
Explicit override — pass agent: "claude" or agent: "codex" to force a runtime regardless of the model id (e.g. agent: "codex" with a non-OpenAI model). Omit agent to auto-select from the model id (default: claude). The benchmark endpoint (POST /api/v1/benchmarks/runs) accepts the same agent field.

​
Claude Agent Models

Used when type: "claude" (the default). Pass these IDs in the model field of your request body. The agent runtime (Claude vs Codex) is auto-selected per the rule above.
Model NameModel IDNotes
Claude Sonnet 4.6blackboxai/anthropic/claude-sonnet-4.6Default — best balance of speed and capability
Blackbox Problackboxai/blackbox-proBlackbox flagship agent model
Claude Opus 4.7blackboxai/anthropic/claude-opus-4.7Latest Opus generation
Claude Opus 4.6blackboxai/anthropic/claude-opus-4.6Most capable, slower
Claude Sonnet 4.5blackboxai/anthropic/claude-sonnet-4.5Previous Sonnet generation
Claude Haiku 4.5blackboxai/anthropic/claude-haiku-4.5Fast, lightweight Claude model
GPT-5.5 Problackboxai/openai/gpt-5.5-proOpenAI top-tier Pro model
GPT-5.5blackboxai/openai/gpt-5.5OpenAI flagship
GPT-5.4 Problackboxai/openai/gpt-5.4-proOpenAI high-capability model
GPT-5.4blackboxai/openai/gpt-5.4OpenAI general-purpose model
GPT-5.4 Nanoblackboxai/openai/gpt-5.4-nanoCompact, low-latency OpenAI model
GPT-5.3 Codexblackboxai/openai/gpt-5.3-codexOpenAI code-specialized model
Nemotron 3 Nano 30B A3Bblackboxai/nvidia/nemotron-3-nano-30b-a3bNVIDIA Nemotron compact MoE
Nemotron Nano 12B v2 VLblackboxai/nvidia/nemotron-nano-12b-v2-vlNVIDIA Nemotron vision-language
Grok 4.3blackboxai/x-ai/grok-4.3xAI flagship
Grok 4.1 Fast (Non-Reasoning)blackboxai/x-ai/grok-4.1-fast-non-reasoningxAI fast non-reasoning model
Grok Build 0.1blackboxai/x-ai/grok-build-0.1xAI build-focused model
MiniMax M2.7blackboxai/minimax/minimax-m2.7Latest MiniMax flagship
MiniMax M2.5blackboxai/minimax/minimax-m2.5MiniMax previous flagship
Gemini 3.5 Flashblackboxai/google/gemini-3.5-flashGoogle Gemini fast multimodal
Gemini 3.1 Pro Previewblackboxai/google/gemini-3.1-pro-previewGoogle Gemini Pro preview
Gemini 3.1 Flash Liteblackboxai/google/gemini-3.1-flash-liteGoogle Gemini lightweight
Gemma 4 26B A4B ITblackboxai/google/gemma-4-26b-a4b-itGoogle Gemma instruction-tuned MoE
Kimi K2.6blackboxai/moonshotai/kimi-k2.6Moonshot AI flagship
Devstral 2blackboxai/mistral/devstral-2Mistral developer model
Ministral 8Bblackboxai/mistral/ministral-8bMistral compact 8B model
Ministral 3Bblackboxai/mistral/ministral-3bMistral compact 3B model
Pixtral Largeblackboxai/mistral/pixtral-largeMistral large multimodal model
Pixtral 12Bblackboxai/mistral/pixtral-12bMistral multimodal model
Codestralblackboxai/mistral/codestralMistral code-specialized model
Mistral Medium 3.5blackboxai/mistral/mistral-medium-3.5Mistral medium-tier model
Mistral Smallblackboxai/mistral/mistral-smallMistral small general-purpose model
Mistral Nemoblackboxai/mistral/mistral-nemoMistral Nemo model
Llama 3.2 3Bblackboxai/meta/llama-3.2-3bMeta Llama compact
Llama 3.2 1Bblackboxai/meta/llama-3.2-1bMeta Llama smallest
Llama 3.1 70Bblackboxai/meta/llama-3.1-70bMeta Llama 3.1 large
Llama 3.1 8Bblackboxai/meta/llama-3.1-8bMeta Llama 3.1
Morph v3 Largeblackboxai/morph/morph-v3-largeMorph large model
Morph v3 Fastblackboxai/morph/morph-v3-fastMorph fast model
Nova 2 Liteblackboxai/amazon/nova-2-liteAmazon Nova lightweight model
Nova Microblackboxai/amazon/nova-microAmazon Nova smallest model
DeepSeek V4 Flashblackboxai/deepseek/deepseek-v4-flashDeepSeek fast model
MiMo v2.5 Problackboxai/xiaomi/mimo-v2.5-proXiaomi MiMo flagship
MiMo v2.5blackboxai/xiaomi/mimo-v2.5Xiaomi MiMo base model
Qwen 3.7 Maxblackboxai/alibaba/qwen3.7-maxAlibaba Qwen next-gen flagship
Qwen 3.6 Max Previewblackboxai/alibaba/qwen3.6-max-previewAlibaba Qwen flagship preview
Qwen 3.6 Plusblackboxai/alibaba/qwen3.6-plusAlibaba Qwen Plus tier
Qwen 3.6 27Bblackboxai/alibaba/qwen3.6-27bAlibaba Qwen mid-size
Qwen3 VL 235B A22B Instructblackboxai/alibaba/qwen3-vl-235b-a22b-instructQwen3 vision-language MoE
Qwen3 Max Thinkingblackboxai/alibaba/qwen3-max-thinkingQwen3 reasoning-focused
Qwen3 Maxblackboxai/qwen3-maxQwen3 flagship
Qwen3 Coderblackboxai/qwen3-coderQwen3 code-specialized
Qwen3 VL 235B A22Bblackboxai/qwen3-vl-235b-a22bQwen3 vision-language MoE
Qwen3 VL 32Bblackboxai/qwen3-vl-32bQwen3 vision-language 32B
GLM-5blackboxai/z-ai/glm-5Z.AI GLM flagship
GLM-5.1blackboxai/z-ai/glm-5.1Z.AI GLM next iteration
GLM-5V Turboblackboxai/z-ai/glm-5v-turboZ.AI GLM vision turbo
GLM-4.7 Flashblackboxai/z-ai/glm-4.7-flashZ.AI GLM fast model
Trinity Large Thinkingblackboxai/arcee-ai/trinity-large-thinkingArcee reasoning model
KAT Coder Pro v2blackboxai/kwaipilot/kat-coder-pro-v2Kwaipilot code model
If you omit the model field, the API defaults to blackboxai/anthropic/claude-sonnet-4.6.

​
Example — Single Agent with Model Override

curl -X POST 'https://agent.blackbox.ai/api/v1/tasks' \
  -H 'Authorization: Bearer YOUR_API_KEY' \
  -H 'Content-Type: application/json' \
  -d '{
    "prompt": "Refactor the auth module to use middleware",
    "model": "blackboxai/anthropic/claude-opus-4.7"
  }'

​
Standard Chat Models

Used when type: "standard". Pass these IDs in the model field.
Model NameModel IDNotes
Mistral Smallmistral/mistral-smallDefault — fast, streaming, tool use
MiniMax M2.5minimax/minimax-m2.5MiniMax flagship
Kimi K2.5moonshotai/kimi-k2.5Moonshot AI flagship
Grok 4.1 Fastxai/grok-4.1-fast-non-reasoningxAI fast non-reasoning model
If you omit the model field in standard mode, the API defaults to mistral/mistral-small.

​
Example — Standard Chat with Model Override

curl -X POST 'https://agent.blackbox.ai/api/v1/tasks' \
  -H 'Authorization: Bearer YOUR_API_KEY' \
  -H 'Content-Type: application/json' \
  -d '{
    "prompt": "Explain async/await in JavaScript",
    "type": "standard",
    "model": "moonshotai/kimi-k2.5"
  }'