Claude Opus 5 benchmark ranking on every AI leaderboard
closed weights · #3 overall, podium score 38.7 · #1 on 2 boards · top 10 on 10 of 10 · released Jul 24, 2026
10 of 10 boards
| Rank | Board | Score | |||
|---|---|---|---|---|---|
| 1 |  SWE-Bench ProCoding · as “Opus 5 (Claude Code) xhigh” | Coding | 98.0% | 11 | |
| 1 |  Document ArenaVision · as “claude-opus-5-high” | Vision | 1,516 | 38 | |
| 3 | MCP AtlasAgents and tools · as “claude-opus-5 (xhigh)” | Agents and tools | 85.8% | 34 | |
| 4 |  Epoch ECIReasoning · as “Claude Opus 5” | Reasoning | 162.7 | 100 | |
| 4 |  ARC-AGI-2Reasoning · as “Claude Opus 5 (Max)” | Reasoning | 90.4% | 78 | |
| 4 | WebDev ArenaCoding · as “claude-opus-5-max” | Coding | 1,692 | 100 | |
| 8 | Text ArenaChat · as “claude-opus-5-high” | Chat | 1,493 | 100 | |
| 8 | Vision ArenaVision · as “claude-opus-5-high” | Vision | 1,289 | 100 | |
| 9 |  LiveBenchReasoning · as “claude-opus-5-max-effort” | Reasoning | 80.1% | 59 | |
| 10 | GPQA DiamondReasoning · as “claude-opus-5_max” | Reasoning | 93.9% | 100 |
Category ranks and more models
Scores as published by each board on the capture date. Model names and logos belong to their owners; logos via logo.dev.