B.O.B. Leaderboard
Anthropicreleased 22 Sep 2026
Claude Opus 5.5
#1 of 9531 of 46 benchmarksConsistent
95.9
B.O.B.
Category profile
- Real use (hard to game)2×97
- Reasoning & knowledge90
- Math96
- Coding98
- Agents & real work94
- Independent indexes100
Reality check
Lab tests
94
Real use (7 tests)
97
Lab tests and real-use tests agree. It does 2 points better on tests that are hard to train for than on the usual lab tests.
What counts as real usePrice per 1M tokens (OpenRouter)
$4.00 in · $20.00 out
Cost per task on the AA index
$5.98
Output speed (Artificial Analysis)
95 tok/s
§Receipts
Every benchmark result
“Beats” = the share of tracked models this one beat on that test. The raw score is its best reported setting.
| Benchmark | Raw | Beats | Source |
|---|---|---|---|
| Real use (hard to game) · Real use 97 | |||
| BridgeBench (overall)as Claude Opus 5.5 | 760 | 100%of 29 | BridgeMind |
| LMArena Agent Arenaas Claude Opus 5.5 (Max) | 0.170 | 100%of 50 | LMArena (real agent sessions) |
| LMArena WebDevas claude-opus-5.5-max | 1811 | 100%of 82 | LMArena (blind human votes) |
| LMArena Textas claude-opus-5.5-high | 1508 | 98%of 80 | LMArena (blind human votes) |
| LiveBench 2.0 (real-work suites)as claude-opus-5.5 | 86.6% | 97%of 39 | LiveBench / Abacus.AI |
| Vals Indexas anthropic/claude-opus-5-5 | 67.0% | 95%of 42 | Vals AI (private test sets) |
| EQ-Bench Creative Writingas claude-opus-5-5 | 2050 | 87%of 46 | Sam Paech (open source) |
| Reasoning & knowledge · Reasoning 90 | |||
| CritPt (physics research)as claude-opus-5-5_max | 31.7% | 98%of 89 | Epoch AI Benchmarking Hub |
| Blueprint-Bench 2as claude-opus-5-5_unknown | 51.2% | 97%of 31 | Epoch AI Benchmarking Hub |
| Mystery Game Puzzlesas claude-opus-5-5_max | 71.0% | 96%of 57 | Epoch AI Benchmarking Hub |
| Chess Puzzlesas claude-opus-5-5_max | 67.0% | 96%of 73 | Epoch AI Benchmarking Hub |
| ARC-AGI-2as claude-opus-5-5_high | 93.3% | 95%of 44 | Epoch AI Benchmarking Hub |
| SimpleQA Verifiedas claude-opus-5-5_max | 72.2% | 93%of 57 | Epoch AI Benchmarking Hub |
| GPQA Diamondas claude-opus-5-5_max | 90.6% | 57%of 70 | Epoch AI Benchmarking Hub |
| Math · Math 96 | |||
| ProofBenchas claude-opus-5-5_max | 100.0% | 99%of 68 | Epoch AI Benchmarking Hub |
| FrontierMath Tiers 1-3as claude-opus-5-5_max | 91.2% | 97%of 60 | Epoch AI Benchmarking Hub |
| FrontierMath Tier 4as claude-opus-5-5_max | 95.0% | 96%of 52 | Epoch AI Benchmarking Hub |
| OTIS Mock AIMEas claude-opus-5-5_max | 100.0% | 94%of 73 | Epoch AI Benchmarking Hub |
| Coding · Coding 98 | |||
| FrontierCodeas claude-opus-5-5_medium | 54.6% | 100%of 45 | Epoch AI Benchmarking Hub |
| CursorBenchas claude-opus-5-5_max | 57.8% | 100%of 17 | Epoch AI Benchmarking Hub |
| SciCodeas claude-opus-5-5_max | 66.9% | 100%of 86 | Epoch AI Benchmarking Hub |
| FrontierSWEas claude-opus-5-5_max | 62.3% | 95%of 20 | Epoch AI Benchmarking Hub |
| ALE-Benchas claude-opus-5-5_high | 2147.4% | 93%of 71 | Epoch AI Benchmarking Hub |
| Agents & real work · Agents 94 | |||
| Remote Labor Indexas claude-opus-5-5_unknown | 21.3% | 100%of 9 | Epoch AI Benchmarking Hub |
| DTBenchas claude-opus-5-5_max | 98.9% | 100%of 67 | Epoch AI Benchmarking Hub |
| LMCAas claude-opus-5-5_max | 68.2% | 100%of 67 | Epoch AI Benchmarking Hub |
| APEX-Agentsas claude-opus-5-5_max | 73.5% | 96%of 48 | Epoch AI Benchmarking Hub |
| GDP.pdfas claude-opus-5-5_max | 30.6% | 86%of 37 | Epoch AI Benchmarking Hub |
| Vending-Bench 2as claude-opus-5-5_unknown | 9235.2% | 80%of 47 | Epoch AI Benchmarking Hub |
| Independent indexes · Indexes 100 | |||
| Epoch Capabilities Indexas Claude Opus 5.5 | 166.8 | 100%of 81 | Epoch AI |
| Artificial Analysis Intelligence Indexas Claude Opus 5.5 (max with fallback) | 58 | 100%of 54 | Artificial Analysis |