LLM Router Model Scope

Login-free access to the smart-routing leaderboard: model scores across 5 categories and 23 sub-dimensions, plus the routable model pool grouped by vendor

GET /api/router/leaderboard

LLM Router Model Scope

cURL

curl --request GET \
  --url https://aihubmix.com/api/router/leaderboard \
  --header 'Authorization: Bearer <token>'

import requests

url = "https://aihubmix.com/api/router/leaderboard"

headers = {"Authorization": "Bearer <token>"}

response = requests.get(url, headers=headers)

print(response.text)

200

{
  "dims": {
    "text.coding": [
      { "m": "claude-opus-4-6", "v": "anthropic", "s": 85.4, "pin": 2.5, "pout": 5, "lat": 822 }
    ]
  },
  "pool": {
    "anthropic": ["claude-opus-4-6"]
  },
  "poolCount": 37
}

Request Parameters

This is a login-free open endpoint — no authentication required; it can be called directly from frontend pages or third parties. CORS is enabled and the global API rate limit applies. No request parameters.

Response Parameters

The response is a raw JSON object (no success / data wrapper), aligned with the frontend leaderboard data structure.

dims · object

Per-dimension leaderboards. Keys are dimension names (5 categories, 23 sub-dimensions — see the table below); values are arrays of models in descending score order. Only models that are currently routable appear; manually excluded models are omitted. When a model has no dedicated score for a sub-dimension, it still participates via the dimension fallback chain (e.g. text.coding → text.overall), using the parent dimension’s score.

m · string

Model name.

v · string

Model vendor id (lowercase). Usable as a key into the Model Vendor Icons response object; that endpoint is cached for 5 minutes, so a newly pooled vendor may be briefly missing — fall back gracefully on absent keys.

s · number

Normalized score (0–100, one decimal place). When the model has no dedicated score for the sub-dimension, this is normalized from the fallback parent dimension’s score.

pin · number

Input price coefficient. This is the platform’s raw billing coefficient, not a USD unit price — suitable only for relative comparison between models; for USD prices, query the Model Management API.

pout · number

Output price multiplier, relative to the input price — not a standalone price. A model’s output price coefficient = pin × pout; do not compare pout across models directly as an output price.

lat · integer

First-token latency in milliseconds. null when no measured data is available.

pool · object

The routable model pool. Keys are vendor ids; values are the vendor’s currently pooled model names (alphabetical).

poolCount · integer

Total number of pooled models (deduplicated).

Dimensions

Category	Sub-dimensions
Text (11)	text.overall, text.coding, text.math, text.instruction_following, text.longer_query, text.language.chinese, text.language.english, text.occupational.software_it, text.occupational.mathematical, text.occupational.legal, text.occupational.medicine
Vision (6)	vision.overall, vision.ocr, vision.diagram, vision.homework, vision.chinese, vision.english
Web Dev (4)	code_webdev.overall, code_webdev.html, code_webdev.react, code_image_to_webdev.overall
Search (1)	search.overall
Document (1)	document.overall

These 23 sub-dimensions are the publicly surfaced subset of the 30+ routing dimensions built into the LLM Router; image/video generation dimensions are not included.

200

{
  "dims": {
    "text.coding": [
      { "m": "claude-opus-4-6", "v": "anthropic", "s": 85.4, "pin": 2.5, "pout": 5, "lat": 822 }
    ]
  },
  "pool": {
    "anthropic": ["claude-opus-4-6"]
  },
  "poolCount": 37
}

---

Last updated: 2026-07-13