B.O.B. Leaderboard
Anthropicreleased 28 Sep 2026
Claude Sonnet 5.5
#6 of 9527 of 46 benchmarksConsistent
90.6
B.O.B.
Category profile
- Real use (hard to game)2×89
- Reasoning & knowledge84
- Math92
- Coding91
- Agents & real work95
- Independent indexes96
Reality check
Lab tests
90
Real use (6 tests)
89
Lab tests and real-use tests agree. Its lab-test and real-use scores are within 2 points of each other.
What counts as real usePrice per 1M tokens (OpenRouter)
$2.00 in · $10.00 out
Cost per task on the AA index
$5.46
Output speed (Artificial Analysis)
135 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 89 | |||
| Vals Indexas anthropic/claude-sonnet-5-5 | 67.0% | 98%of 42 | Vals AI (private test sets) |
| LMArena WebDevas claude-sonnet-5.5-xhigh | 1774 | 98%of 82 | LMArena (blind human votes) |
| LiveBench 2.0 (real-work suites)as claude-sonnet-5.5 | 85.2% | 92%of 39 | LiveBench / Abacus.AI |
| LMArena Agent Arenaas Claude Sonnet 5.5 (Max) | 0.130 | 90%of 50 | LMArena (real agent sessions) |
| BridgeBench (overall)as Claude Sonnet 5.5 | 647 | 89%of 29 | BridgeMind |
| LMArena Textas claude-sonnet-5.5-xhigh | 1475 | 66%of 80 | LMArena (blind human votes) |
| Reasoning & knowledge · Reasoning 84 | |||
| GPQA Diamondas claude-sonnet-5-5_max | 95.6% | 97%of 70 | Epoch AI Benchmarking Hub |
| CritPt (physics research)as claude-sonnet-5-5_max | 31.4% | 96%of 89 | Epoch AI Benchmarking Hub |
| Mystery Game Puzzlesas claude-sonnet-5-5_max | 65.0% | 95%of 57 | Epoch AI Benchmarking Hub |
| Chess Puzzlesas claude-sonnet-5-5_max | 59.0% | 90%of 73 | Epoch AI Benchmarking Hub |
| Blueprint-Bench 2as claude-sonnet-5-5_unknown | 39.3% | 87%of 31 | Epoch AI Benchmarking Hub |
| Humanity's Last Examas claude-sonnet-5-5_xhigh | 46.1% | 75%of 13 | Epoch AI Benchmarking Hub |
| SimpleQA Verifiedas claude-sonnet-5-5_max | 46.5% | 50%of 57 | Epoch AI Benchmarking Hub |
| Math · Math 92 | |||
| ProofBenchas claude-sonnet-5-5_max | 100.0% | 99%of 68 | Epoch AI Benchmarking Hub |
| OTIS Mock AIMEas claude-sonnet-5-5_max | 100.0% | 94%of 73 | Epoch AI Benchmarking Hub |
| FrontierMath Tiers 1-3as claude-sonnet-5-5_max | 88.8% | 90%of 60 | Epoch AI Benchmarking Hub |
| FrontierMath Tier 4as claude-sonnet-5-5_max | 80.5% | 85%of 52 | Epoch AI Benchmarking Hub |
| Coding · Coding 91 | |||
| SciCodeas claude-sonnet-5-5_max | 61.0% | 96%of 86 | Epoch AI Benchmarking Hub |
| CursorBenchas claude-sonnet-5-5_max | 55.5% | 94%of 17 | Epoch AI Benchmarking Hub |
| FrontierCodeas claude-sonnet-5-5_xhigh | 52.1% | 91%of 45 | Epoch AI Benchmarking Hub |
| FrontierSWEas claude-sonnet-5-5_max | 61.9% | 90%of 20 | Epoch AI Benchmarking Hub |
| ALE-Benchas claude-sonnet-5-5_high | 1819.1% | 86%of 71 | Epoch AI Benchmarking Hub |
| Agents & real work · Agents 95 | |||
| APEX-Agentsas claude-sonnet-5-5_max | 75.5% | 98%of 48 | Epoch AI Benchmarking Hub |
| GDP.pdfas claude-sonnet-5-5_max | 31.4% | 94%of 37 | Epoch AI Benchmarking Hub |
| Vending-Bench 2as claude-sonnet-5-5_unknown | 11165.3% | 91%of 47 | Epoch AI Benchmarking Hub |
| Independent indexes · Indexes 96 | |||
| Artificial Analysis Intelligence Indexas Claude Sonnet 5.5 (max with fallback) | 56 | 97%of 54 | Artificial Analysis |
| Epoch Capabilities Indexas Claude Sonnet 5.5 | 164.4 | 94%of 81 | Epoch AI |