B.O.B. Leaderboard
Google DeepMindreleased 13 Aug 2026
Gemini 3.7 Flash
#22 of 9527 of 46 benchmarksConsistent
62.4
B.O.B.
Category profile
- Real use (hard to game)2×51
- Reasoning & knowledge76
- Math61
- Coding55
- Agents & real work64
- Independent indexes79
Reality check
Lab tests
64
Real use (6 tests)
51
Lab tests and real-use tests agree. It does 12 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)
$0.75 in · $3.75 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 51 | |||
| LMArena Textas gemini-3.7-flash-high | 1487 | 82%of 80 | LMArena (blind human votes) |
| LMArena WebDevas gemini-3.7-flash-high | 1590 | 69%of 82 | LMArena (blind human votes) |
| EQ-Bench Creative Writingas gemini-3.7-flash | 1723 | 42%of 46 | Sam Paech (open source) |
| Vals Indexas google/gemini-3.7-flash | 51.3% | 42%of 42 | Vals AI (private test sets) |
| LiveBench 2.0 (real-work suites)as gemini-3.7-flash | 62.9% | 37%of 39 | LiveBench / Abacus.AI |
| LMArena Agent Arenaas Gemini 3.7 Flash (High) | -0.005 | 37%of 50 | LMArena (real agent sessions) |
| Reasoning & knowledge · Reasoning 76 | |||
| GPQA Diamondas gemini-3.7-flash_high | 94.8% | 93%of 70 | Epoch AI Benchmarking Hub |
| SimpleQA Verifiedas gemini-3.7-flash_high | 69.2% | 84%of 57 | Epoch AI Benchmarking Hub |
| ARC-AGI-2as gemini-3.7-flash_high | 84.6% | 77%of 44 | Epoch AI Benchmarking Hub |
| Chess Puzzlesas gemini-3.7-flash_high | 47.0% | 74%of 73 | Epoch AI Benchmarking Hub |
| Mystery Game Puzzlesas gemini-3.7-flash_high | 37.0% | 72%of 57 | Epoch AI Benchmarking Hub |
| CritPt (physics research)as gemini-3.7-flash_high | 14.3% | 55%of 89 | Epoch AI Benchmarking Hub |
| Math · Math 61 | |||
| ProofBenchas gemini-3.7-flash_unknown | 58.0% | 67%of 68 | Epoch AI Benchmarking Hub |
| FrontierMath Tiers 1-3as gemini-3.7-flash_high | 71.6% | 63%of 60 | Epoch AI Benchmarking Hub |
| OTIS Mock AIMEas gemini-3.7-flash_high | 97.2% | 62%of 73 | Epoch AI Benchmarking Hub |
| FrontierMath Tier 4as gemini-3.7-flash_high | 36.6% | 53%of 52 | Epoch AI Benchmarking Hub |
| Coding · Coding 55 | |||
| SciCodeas gemini-3.7-flash_medium | 59.8% | 92%of 86 | Epoch AI Benchmarking Hub |
| FrontierCodeas gemini-3.7-flash_unknown | 43.6% | 61%of 45 | Epoch AI Benchmarking Hub |
| DeepSWEas gemini-3.7-flash_medium | 65.5% | 56%of 26 | Epoch AI Benchmarking Hub |
| ALE-Benchas gemini-3.7-flash_high | 904.3% | 37%of 71 | Epoch AI Benchmarking Hub |
| FrontierSWEas gemini-3.7-flash_high | 20.3% | 26%of 20 | Epoch AI Benchmarking Hub |
| Agents & real work · Agents 64 | |||
| APEX-Agentsas gemini-3.7-flash_unknown | 67.8% | 92%of 48 | Epoch AI Benchmarking Hub |
| DTBenchas gemini-3.7-flash_high | 96.8% | 88%of 67 | Epoch AI Benchmarking Hub |
| LMCAas gemini-3.7-flash_high | 50.4% | 70%of 67 | Epoch AI Benchmarking Hub |
| GDP.pdfas gemini-3.7-flash_high | 23.8% | 58%of 37 | Epoch AI Benchmarking Hub |
| Remote Labor Indexas gemini-3.7-flash_unknown | 5.0% | 13%of 9 | Epoch AI Benchmarking Hub |
| Independent indexes · Indexes 79 | |||
| Epoch Capabilities Indexas Gemini 3.7 Flash | 157.2 | 79%of 81 | Epoch AI |