Models

# Models

BrainstormRouter provides access to 362 model endpoints across 20+ providers through a single API. Every model is available via the standard OpenAI-compatible chat completions endpoint.

Featured Models

Anthropic

| Model | Context | Strengths |

|-------|---------|-----------|

| Claude Opus 4.6 | 200K | Top-tier reasoning, complex code, architecture |

| Claude Sonnet 4.6 | 200K | Balanced quality and speed, everyday coding |

| Claude Haiku 3.5 | 200K | Fast responses, cost-efficient for simple tasks |

OpenAI

| Model | Context | Strengths |

|-------|---------|-----------|

| GPT-5.4 | 128K | Strong general-purpose, creative writing |

| GPT-4.1 | 128K | Reliable workhorse, broad capabilities |

| o3 | 128K | Advanced reasoning and math |

Google

| Model | Context | Strengths |

|-------|---------|-----------|

| Gemini 3.1 Pro | 2M | Largest context window, multimodal |

| Gemini 3.1 Flash | 1M | Fastest response times, great for bulk work |

DeepSeek

| Model | Context | Strengths |

|-------|---------|-----------|

| DeepSeek V3 | 128K | Cost-efficient coding, strong at Python |

| DeepSeek R1 | 128K | Reasoning-focused, math and logic |

Other Providers

BrainstormRouter also routes to models from: Groq, Mistral, xAI (Grok), Together AI, Fireworks, Anyscale, Perplexity, Cohere, AI21, Replicate, Lepton, Cerebras, SambaNova, OctoAI, Modal, Baseten, and more.

Model Selection

Auto Mode

Set model: "auto" and BrainstormRouter selects the best model for each request using the [13 intelligence systems](/router/intelligence):

``json

{"model": "auto", "messages": [...]}

`

Direct Selection

Specify a model directly to bypass routing:

`json

{"model": "claude-opus-4-6", "messages": [...]}

`

Strategy Override

Use the X-BR-Strategy header to control how auto mode selects:

• quality -- always pick the highest-quality model
• cost -- minimize cost above a quality floor
• combined -- weighted balance (default)
• capability -- match on required features (tools, vision, long context)

Listing Models

`bash

curl https://api.brainstormrouter.com/v1/models \

-H "Authorization: Bearer br-your-api-key"

`

Each model entry includes:

• id -- the model identifier to use in requests
• provider -- which provider serves this model
• context_window -- maximum token context
• pricing -- input and output cost per million tokens
• capabilities -- supported features (streaming, tools, vision, JSON mode)
• status -- current availability (online, degraded, offline)

Real-Time Availability

Model availability is updated in real-time based on provider health checks. The /v1/intelligence/health endpoint shows current status for all providers with latency percentiles and error rates.

When a model is degraded or offline, BrainstormRouter automatically falls back to the next best alternative without client-side changes.

Custom Model Aliases

Enterprise plans support custom model aliases for organizational standards:

`json

{"model": "our-default", "messages": [...]}

``

Aliases map to routing rules configured in the dashboard, letting you change the underlying model without updating client code.