Models

List available models from the unified catalog.

List models

/v1/models

Returns available models for the authenticated tenant. Respects API key model allowlists.

Response

{
  "object": "list",
  "data": [
    {
      "id": "anthropic/claude-sonnet-4",
      "object": "model",
      "created": 1700000000,
      "owned_by": "anthropic"
    },
    {
      "id": "openai/gpt-4o",
      "object": "model",
      "created": 1700000000,
      "owned_by": "openai"
    }
  ]
}

Authentication

Requires API key: Authorization: Bearer br_live_...

Model naming convention

All models use the provider/model format. Examples:

Model ID	Provider	Description
`anthropic/claude-sonnet-4`	Anthropic	Latest Sonnet
`anthropic/claude-haiku-4-5`	Anthropic	Fast, cheap
`openai/gpt-4o`	OpenAI	Best GPT
`openai/gpt-4o-mini`	OpenAI	Fast, cheap
`google/gemini-2.0-flash`	Google	Fast multimodal
`groq/llama-3.3-70b`	Groq	Fast open-source

Variants

Append a suffix for routing behavior:

anthropic/claude-sonnet-4:floor — cheapest endpoint
anthropic/claude-sonnet-4:fast — lowest latency
anthropic/claude-sonnet-4:best — highest quality
anthropic/claude-sonnet-4:smart — Thompson sampling

Presets

Use @preset/slug as the model name to reference a saved routing preset.

curl https://api.brainstormrouter.com/v1/models \
  -H "Authorization: Bearer br_live_..."

Model leaderboard

/v1/models/leaderboard

Returns model performance rankings powered by Thompson sampling reward scores. Data is aggregated from a 7-day rolling window of real traffic.

Query parameters

Param	Type	Default	Description
`sort`	string	`overall`	Sort dimension: `overall`, `quality`, `speed`, `value`, `reliability`
`provider`	string	—	Filter to a specific provider (e.g., `anthropic`)
`limit`	integer	`100`	Max results (capped at 500)
`offset`	integer	`0`	Pagination offset

Response

{
  "object": "list",
  "sort_by": "overall",
  "model_count": 42,
  "cached_at": "2026-02-26T12:00:00.000Z",
  "data": [
    {
      "id": "anthropic/claude-sonnet-4",
      "provider": "anthropic",
      "model_id": "claude-sonnet-4",
      "reward_score": 0.8523,
      "reward_variance": 0.0012,
      "validity": 0.9501,
      "quality": 0.91,
      "latency_ms": 520.3,
      "tokens_per_second": 85.2,
      "success_rate": 0.998,
      "pricing": { "input": 3, "output": 15 },
      "capabilities": ["streaming", "tools", "vision"],
      "sample_count": 12430,
      "days_active": 7,
      "error_rate": 0.002,
      "value_score": 284.1
    }
  ]
}

Sort dimensions

Dimension	Description
`overall`	Thompson sampling composite reward (latency + cost + quality + validity)
`quality`	Tool-call accuracy (JSON validity 30% + schema compliance 40% + execution 30%)
`speed`	Lowest EWMA latency first
`value`	Reward per dollar — best bang for buck
`reliability`	Highest success rate first

SDK example

const leaderboard = await client.models.leaderboard({
  sort: "value",
  provider: "anthropic",
});

for (const model of leaderboard.data) {
  console.log(`${model.id}: reward=${model.reward_score} value=${model.value_score}`);
}

curl "https://api.brainstormrouter.com/v1/models/leaderboard?sort=overall&limit=10" \
  -H "Authorization: Bearer br_live_..."