Leaderboard data is updated multiple times per week. Subscribe to the mailing list at lightsage.com/research for updates delivered directly to your inbox.
May 12, 2026
Updated the Claude Code API leaderboard with the latest evaluation runs across 85 companies. Firecrawl holds #1 with a score of 86; OpenRouter and Jina AI are tied at #2 with 84.Top 10 Overall (API leaderboard)
May 9, 2026
Updated the Codex MCP leaderboard with gpt-5.5 runs across 68 companies. Jina AI holds #1 with 85; Stripe enters at #2 with 83. Added 8 new companies including Telnyx, Twilio, Resemble AI, Pipecat, and Fireworks AI.New Companies
Top 10 Overall (Codex MCP leaderboard)
May 8, 2026
Launched the Codex MCP leaderboard with 53 companies scored. Added Voice Telephony as a new category with Bandwidth and Infobip. Jina AI leads at #1 with a score of 81; Firecrawl and Agentmail round out the top 3.New Categories
- Voice Telephony — Telephony and programmable voice call APIs
New Companies
Top 10 Overall (Codex MCP leaderboard, launch)
April 30, 2026
Updated the Claude Code API, Claude Code CLI, Codex API, and Codex CLI leaderboards. This snapshot reflects the latest completed runs with 108 companies listed and 954 completed evaluations across all four leaderboards.Top 10 Overall (all leaderboards, best score per company)
April 29, 2026
Updated the Codex CLI leaderboard. This snapshot reflects the latest completed Codex CLI runs with 88 companies listed and 216 completed Codex CLI evaluations.Top 10 Overall (Codex CLI leaderboard)
April 26, 2026
Added the Codex CLI leaderboard. This snapshot reflects the initial completed Codex CLI end-to-end runs with 74 companies listed and 33 completed Codex CLI evaluations.Top 10 Overall (Codex CLI leaderboard, launch)
April 24, 2026
Added the Codex API leaderboard. This snapshot reflects the initial completed Codex API runs with 74 companies evaluated.Top 10 Overall (Codex API leaderboard, launch)
April 15, 2026
Added the MCP leaderboard. This snapshot reflects the initial completed MCP runs with 74 companies on the leaderboard and 50 companies tracked with MCP metadata in the registry.Top 10 Overall (MCP leaderboard, launch)
April 9, 2026
Added the Claude Code CLI leaderboard. This snapshot reflects the initial completed Claude Code CLI end-to-end runs with 74 companies on the leaderboard and 35 companies tracked with CLI metadata in the registry.Top 10 Overall (Claude Code CLI leaderboard, launch)
April 5, 2026
Added 2 new categories to cover APIs that did not fit existing buckets, along with Datadog and Agentmail as the first entries in each.New Categories
- Observability — Monitoring, metrics, and event tracking APIs
- Email — Programmatic email inbox and messaging APIs
New Companies
Top 10 Overall (Claude Code API leaderboard)
March 2026
Launched the Devtool Arena leaderboard with 72 API companies scored across 13 initial categories.Initial Launch Categories and Companies
Inference (8 companies)
Inference (8 companies)
Groq, TogetherAI, Perplexity, OpenRouter, Cerebras, Fireworks AI, DeepInfra, SambaNova
Auth (6 companies)
Auth (6 companies)
Clerk, Auth0, Stytch, Scalekit, WorkOS, Descope
Payment (10 companies)
Payment (10 companies)
Stripe, PayPal, Square, Adyen, Razorpay, Mollie, Paddle, Chargebee, LemonSqueezy, Airwallex
Search (6 companies)
Search (6 companies)
Exa, Tavily, You.com, Brave Search, Firecrawl, Jina AI
Voice STT (3 companies)
Voice STT (3 companies)
Deepgram, AssemblyAI, OpenAI Whisper
Voice TTS (2 companies)
Voice TTS (2 companies)
ElevenLabs, Rime
Voice Infra (1 company)
Voice Infra (1 company)
LiveKit
Sandboxes (6 companies)
Sandboxes (6 companies)
Vercel, Cloudflare Workers, E2B, Daytona, Modal, Fly.io
Vector Databases (6 companies)
Vector Databases (6 companies)
Chroma, Pinecone, Weaviate, Qdrant, LanceDB, Zilliz Cloud
Stablecoin (6 companies)
Stablecoin (6 companies)
Stripe (Crypto), Coinbase Payments, Circle, Fireblocks, DFNS, Triple-A
Meeting Bot (5 companies)
Meeting Bot (5 companies)
Recall.ai, Meeting BaaS, MeetGeek, Meetstream, CueMeet
Durable Workflow (7 companies)
Durable Workflow (7 companies)
Temporal, Prefect, Cadence, Camunda, Netflix Conductor, AWS Step Functions, Akka
Cloud Hosting (6 companies)
Cloud Hosting (6 companies)
Vercel, Cloudflare, Netlify, Railway, Render, DigitalOcean