B.O.B. Leaderboard
OpenAIreleased 9 Jul 2026
GPT-5.6 Sol
#13 of 9537 of 46 benchmarksConsistent
80.7
B.O.B.
Category profile
- Real use (hard to game)2×71
- Reasoning & knowledge87
- Math89
- Coding76
- Agents & real work83
- Independent indexes88
Reality check
Lab tests
84
Real use (10 tests)
71
Lab tests and real-use tests agree. It does 13 points better on the usual lab tests than on tests that are hard to train for.
What counts as real usePrice per 1M tokens (OpenRouter)
$4.00 in · $20.00 out
Cost per task on the AA index
—
Output speed (Artificial Analysis)
—
§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 71 | |||
| SWE-rebenchas GPT-5.6 Sol [medium] | 62.3% | 85%of 34 | Nebius |
| EQ-Bench Creative Writingas gpt-5.6-sol | 1972 | 82%of 46 | Sam Paech (open source) |
| LMArena Agent Arenaas GPT 5.6 Sol (xHigh) | 0.077 | 82%of 50 | LMArena (real agent sessions) |
| LMArena Textas gpt-5.6-sol-xhigh | 1485 | 80%of 80 | LMArena (blind human votes) |
| Vals Indexas openai/gpt-5.6-sol | 58.0% | 76%of 42 | Vals AI (private test sets) |
| LMArena WebDevas gpt-5.6-sol-xhigh (codex-harness) | 1617 | 74%of 82 | LMArena (blind human votes) |
| BridgeBench (overall)as GPT-5.6 Sol | 600 | 64%of 29 | BridgeMind |
| LiveBench 2.0 (real-work suites)as gpt-5.6-sol | 72.5% | 63%of 39 | LiveBench / Abacus.AI |
| SimpleBenchas gpt-5.6-sol_unknown | 64.8% | 53%of 30 | Epoch AI Benchmarking Hub |
| Kagi LLM Benchmarkas gpt-5-6-sol | 67.0% | 52%of 22 | Kagi |
| Reasoning & knowledge · Reasoning 87 | |||
| CritPt (physics research)as gpt-5.6-sol_max | 32.3% | 100%of 89 | Epoch AI Benchmarking Hub |
| ARC-AGI-2as gpt-5.6-sol_max | 92.5% | 93%of 44 | Epoch AI Benchmarking Hub |
| Chess Puzzlesas gpt-5.6-sol_max | 59.0% | 90%of 73 | Epoch AI Benchmarking Hub |
| EnigmaEvalas gpt-5.6-sol_high | 37.1% | 89%of 10 | Epoch AI Benchmarking Hub |
| Mystery Game Puzzlesas gpt-5.6-sol_max | 58.0% | 88%of 57 | Epoch AI Benchmarking Hub |
| SimpleQA Verifiedas gpt-5.6-sol_max | 69.7% | 87%of 57 | Epoch AI Benchmarking Hub |
| GPQA Diamondas gpt-5.6-sol_max | 93.5% | 83%of 70 | Epoch AI Benchmarking Hub |
| Blueprint-Bench 2as gpt-5.6-sol_unknown | 33.6% | 68%of 31 | Epoch AI Benchmarking Hub |
| Math · Math 89 | |||
| OTIS Mock AIMEas gpt-5.6-sol_max | 100.0% | 94%of 73 | Epoch AI Benchmarking Hub |
| FrontierMath Tiers 1-3as gpt-5.6-sol_max | 89.1% | 92%of 60 | Epoch AI Benchmarking Hub |
| FrontierMath Tier 4as gpt-5.6-sol_max | 82.9% | 88%of 52 | Epoch AI Benchmarking Hub |
| ProofBenchas gpt-5.6-sol_max | 83.0% | 83%of 68 | Epoch AI Benchmarking Hub |
| Coding · Coding 76 | |||
| ALE-Benchas gpt-5.6-sol_max | 2176.9% | 96%of 71 | Epoch AI Benchmarking Hub |
| WeirdMLas gpt-5.6-sol_high | 88.8% | 88%of 52 | Epoch AI Benchmarking Hub |
| DeepSWEas gpt-5.6-sol_max | 72.7% | 88%of 26 | Epoch AI Benchmarking Hub |
| SciCodeas gpt-5.6-sol_high | 57.8% | 83%of 86 | Epoch AI Benchmarking Hub |
| FrontierCodeas gpt-5.6-sol_unknown | 47.5% | 73%of 45 | Epoch AI Benchmarking Hub |
| GSO (code optimisation)as gpt-5.6-sol_unknown | 76.5% | 70%of 11 | Epoch AI Benchmarking Hub |
| FrontierSWEas gpt-5.6-sol_max | 32.2% | 58%of 20 | Epoch AI Benchmarking Hub |
| CursorBenchas gpt-5.6-sol_max | 41.7% | 50%of 17 | Epoch AI Benchmarking Hub |
| Agents & real work · Agents 83 | |||
| LMCAas gpt-5.6-sol_xhigh | 58.5% | 89%of 67 | Epoch AI Benchmarking Hub |
| GDP.pdfas gpt-5.6-sol_unknown | 30.7% | 89%of 37 | Epoch AI Benchmarking Hub |
| OSWorld 2.0as gpt-5.6-sol_max | 27.3% | 88%of 9 | Epoch AI Benchmarking Hub |
| Vending-Bench 2as gpt-5.6-sol_unknown | 9619.4% | 85%of 47 | Epoch AI Benchmarking Hub |
| DTBenchas gpt-5.6-sol_xhigh | 95.7% | 77%of 67 | Epoch AI Benchmarking Hub |
| Game Boy Advance evalas gpt-5.6-sol_unknown | 52.6% | 73%of 23 | Epoch AI Benchmarking Hub |
| Independent indexes · Indexes 88 | |||
| Epoch Capabilities Indexas GPT-5.6 Sol | 161.7 | 88%of 81 | Epoch AI |