Artificial Analysis
Artificial Analysis MCP: ivory serif type beside a sculptural purple Artificial Analysis emblem on black.
Ask your assistant about models using Artificial Analysis's independent benchmarks instead of what it remembers from training.
- Language models: the Artificial Analysis Intelligence Index and its evaluations, coding and agentic indexes, context window, price per million tokens, and output speed and time to first token.
- API providers: how each provider serves a model on speed, latency and price, with performance over time.
- Media models: arena rankings for text-to-image, image editing, text-to-video, image-to-video, text-to-speech, speech-to-text, speech-to-speech and music.
Every tool is read-only.
Connect
Server URL: https://mcp.artificialanalysis.ai/mcp (Streamable HTTP). There's no API key. Add the URL to your client, sign in with your Artificial Analysis account, choose an organization and click Allow. Calls count against that organization's API quota, and the data you see follows its plan.
Claude
In claude.ai or Claude Desktop, go to Customize > Connectors, click + Add > Add custom connector, and enter the server URL. In Claude Code:
claude mcp add --transport http --scope user artificial-analysis https://mcp.artificialanalysis.ai/mcp
Then run /mcp and sign in.
ChatGPT
Open Plugins, click +, choose Add custom MCP server, enter the server URL and choose OAuth.
Codex
codex mcp add artificial-analysis --url https://mcp.artificialanalysis.ai/mcp
codex mcp login artificial-analysis
Cursor
Install this plugin from the Cursor Marketplace, or add the server to ~/.cursor/mcp.json:
{
"mcpServers": {
"artificial-analysis": {
"url": "https://mcp.artificialanalysis.ai/mcp"
}
}
}
Gemini CLI
gemini extensions install https://github.com/ArtificialAnalysis/artificial-analysis-plugin
Then run /mcp auth artificial-analysis and sign in.
VS Code
Run MCP: Add Server, or add this to .mcp.json:
{
"mcpServers": {
"artificial-analysis": {
"type": "http",
"url": "https://mcp.artificialanalysis.ai/mcp"
}
}
}
Try asking
- Which language models lead the Artificial Analysis Intelligence Index right now?
- Which providers serve gpt-oss-120B fastest, and what do they charge?
- What are the top text-to-video models on the Artificial Analysis Video Arena?
- Which models score above 50 on the Intelligence Index for under $1 per million input tokens?
Tools
| Tool | What it does | Plan |
|---|---|---|
aa_language_models_free | Language models with headline indexes, token prices and median speed | Free |
aa_language_models | Language models with every evaluation, blended prices and percentiles | Pro |
aa_language_model | One language model by slug | Pro |
aa_language_model_performance | A model's performance over 7, 30 or 90 days | Commercial |
aa_language_providers | API providers and the models they serve | Commercial |
aa_language_provider | One API provider by slug | Commercial |
aa_language_provider_performance | A provider's performance over 7, 30 or 90 days | Commercial |
aa_language_provider_measurements | A provider's raw measurements | Commercial |
aa_media_rankings_free | Arena Elo rankings for image, video, speech and music models | Free |
aa_media_rankings | Arena rankings with rank, samples, pricing and categories | Pro |
artificial_analysis_help | Usage, and the plan each tool needs | Any |
aa_api_context, aa_describe_endpoint, aa_describe_fields, aa_describe_schema | Explain the API and its fields, without using quota | Any |
Troubleshooting
| Symptom | Fix |
|---|---|
| Asked to sign in again every hour | Your client didn't request offline_access, so it has no refresh token. Reconnect, or update the client. |
401 invalid_token with an API key configured | The server takes OAuth sign-in only. Remove the Authorization header so the client signs in. |
Privacy and support
The server sends only the endpoint and query your assistant chooses to the Artificial Analysis API, and keeps no conversation content. See the privacy policy and terms of use. For help, email support@artificialanalysis.ai.