lmarena.ai APIlmarena.ai ↗
Access LMArena leaderboard rankings, ratings, and model metadata across text, code, vision, image, video, and search categories via a structured API.
No input parameters required.
curl -X GET 'https://api.parse.bot/scraper/e3e90e15-f9f2-49dc-a1e2-de9030d6d41d/get_leaderboard_overview' \ -H 'X-API-Key: $PARSE_API_KEY'
Fetch the leaderboard overview showing the default ranked models on the main leaderboard page. Returns entries and metadata.
No input parameters required.
{
"type": "object",
"fields": {
"entries": "array of model ranking objects with rank, modelDisplayName, rating, votes, modelOrganization, license, and pricing fields",
"metadata": "object containing leaderboard metadata"
},
"sample": {
"data": {
"entries": [
{
"rank": 1,
"votes": 506,
"rating": 1361.79,
"license": "Proprietary",
"modelUrl": "https://seed.bytedance.com/en/seedance2_0",
"rankLower": 1,
"rankUpper": 1,
"ratingLower": 1340.59,
"ratingUpper": 1382.99,
"releaseType": null,
"contextLength": null,
"pricePerImage": null,
"pricePerSecond": null,
"modelDisplayName": "dreamina-seedance-2.0-720p",
"modelOrganization": "Bytedance",
"inputPricePerMillion": null,
"outputPricePerMillion": null
}
],
"metadata": {}
},
"status": "success"
}
}About the lmarena.ai API
The LMArena API provides structured access to 13 endpoints covering AI model leaderboard rankings across six modalities: text, code, vision, image generation, video generation, and search. Endpoints like get_text_leaderboard and get_top_models_by_category return per-model fields including Elo-style ratings, vote counts, organization, license, context length, and pricing tier — giving developers a programmatic view of community-ranked model performance as tracked by LMArena.
Leaderboard Endpoints by Modality
Each arena has a dedicated endpoint: get_text_leaderboard, get_code_leaderboard, get_vision_leaderboard, get_text_to_image_leaderboard, get_image_edit_leaderboard, get_search_leaderboard, get_text_to_video_leaderboard, and get_image_to_video_leaderboard. Every leaderboard endpoint returns an entries array and an initialModels array. Each entry includes rank, modelDisplayName, rating, votes, modelOrganization, license, and pricing. The text leaderboard additionally exposes contextLength per model. The initialModels array lists models available for direct interaction, with their id, organization, displayName, and capabilities.
Querying and Filtering
get_top_models_by_category accepts a required category parameter (accepted values: text, code, vision, text-to-image, image-edit, search, text-to-video, image-to-video) and an optional limit integer, returning the top N ranked models from that arena sorted by rank. search_leaderboard_models takes a required query string and an optional category, performing case-insensitive matching against modelDisplayName and organization names — useful for checking whether a specific model or vendor appears in any ranked leaderboard.
Cross-Category and Model Discovery
get_best_coding_models returns two arrays in a single call: the full code_leaderboard and a text_leaderboard_snapshot containing the top 10 text arena entries — useful for comparing coding-specific performance against general text generation ranking. list_available_models returns the full set of models exposed for interaction, with per-model inputCapabilities, outputCapabilities, userSelectable, and additional metadata fields, independent of any leaderboard ranking.
Overview Endpoint
get_leaderboard_overview fetches the default main leaderboard page, returning an entries array and a metadata object. This reflects the default ranked view shown on the LMArena homepage without filtering by modality.
- Track how a specific model's Elo rating and vote count evolve across LMArena leaderboard updates.
- Compare a vendor's models across text, code, and vision arenas using
search_leaderboard_modelswith the organization name as the query. - Build a model selection tool that surfaces the top-ranked models per category using
get_top_models_by_categorywith a configurablelimit. - Identify which frontier models support vision or multimodal input by querying
list_available_modelsand inspectinginputCapabilities. - Generate a coding model benchmark digest by combining
get_best_coding_models's code leaderboard with its text arena top-10 snapshot. - Monitor open-weight vs. proprietary model distribution in a leaderboard by filtering
licensefields from any arena endpoint. - Power a leaderboard dashboard that shows rank, rating, votes, and context length for text models alongside pricing tier.
| Tier | Price | Credits/month | Rate limit |
|---|---|---|---|
| Free | $0/mo | 100 | 5 req/min |
| Hobby | $30/mo | 1,000 | 20 req/min |
| Developer | $100/mo | 5,000 | 250 req/min |
One credit = one API call regardless of which marketplace API you call. Exceeding the rate limit returns a 429 response. Authenticate with the X-API-Key header.
Does LMArena offer an official developer API?+
What does the `get_text_leaderboard` endpoint return beyond a model's rank?+
rank, modelDisplayName, rating (the community Elo-style score), votes (total battle count), modelOrganization, license, contextLength, and pricing. The initialModels array in the same response lists models available for interaction with their id, displayName, organization, and capabilities.Can I retrieve historical leaderboard rankings or track rating changes over time?+
Does the API expose individual battle results or head-to-head matchup data between models?+
How does `search_leaderboard_models` behave when no `category` is provided?+
category is omitted, the search runs against the default leaderboard category. Specifying a category (e.g., code, vision, text-to-video) scopes the match to that arena's entries. The response echoes both the query and the resolved category so results are unambiguous.