Google DeepMind·released 2025-02-051 source
Gemini 2.0 Flash (Feb 2025) benchmark scores: 11 benchmarks tracked. 36% of its scores are independently reproduced — the rest are self-reported or unverified.
Leads on 0 of 11 benchmarks·best result 82.17 on MATH level 5 (reproduced)·4 of 11 independently reproduced
| Benchmark | Score | Evidence status | Measured |
|---|---|---|---|
| MATH level 5 | 82.17accuracy (%) | reproduced· optimizedT1 | 2025-02-05Epoch AI |
| GeoBench | 77accuracy (%) | unverifiedT1 | 2025-02-05GeoBench leaderboard |
| Fiction.LiveBench | 61.1accuracy (%) | unverifiedT1 | 2025-02-05Fiction.live leaderboard |
| GPQA diamond | 52.19accuracy (%) | reproduced· optimizedT1 | 2025-02-05Epoch AI |
| DTBench | 38.67accuracy (%) | unverifiedT1 | 2025-02-05https://conceptualreasoning.ai/dtbench |
| OTIS Mock AIME 2024-2025 | 31.04accuracy (%) | reproduced· optimizedT1 | 2025-02-05Epoch AI |
| CadEval | 30accuracy (%) | unverifiedT1 | 2025-02-05CadEval Dashboard |
| WeirdML | 25.77accuracy (%) | unverifiedT1 | 2025-02-05https://htihle.github.io/weirdml.html |
| The Agent Company | 11.4accuracy (%) | unverifiedT1 | 2025-02-05TheAgentCompany experiment results github |
| FrontierMath-2025-02-28-Private | 3.02accuracy (%) | reproduced· optimizedT1 | 2025-02-05Epoch AI |
| ARC-AGI-2 | 1.3accuracy (%) | unverifiedT2 | 2025-02-05 |
Claims drawn from cited facts, not live model generation.
This model was developed by Google DeepMind and released in 2025, placing it among the more recent entrants in the field. With no parameter count disclosed, its scale cannot be quantified, so any assessment of capability versus cost must rest on its origin and vintage rather than on a headline size figure.
3 cited facts
This model is scored on 11 tracked benchmarks, and it holds no current top score on any of them. Because its results have been independently reproduced, outside evaluation backs up the reported scores, so the record can be trusted as verified rather than merely claimed.
3 cited facts
This model trails the best reported score on its benchmarks by an average of 48.6 points, a wide gap that reads as well behind the leaders rather than at or near the frontier. It holds the top score on 0 of its benchmarks, so there is no benchmark among them where it currently stands as the leader.
2 cited facts
1 facts cross-checked across data sources: 1 single-source.
Source facts, citations, and refresh stamp for this record.
Sources: epoch_benchmarks
Gemini 2.0 Flash (Feb 2025) is an AI model developed by Google DeepMind, released 2025-02-05. Its benchmark record below tags every score as independently reproduced, self-reported, or unverified.
Gemini 2.0 Flash (Feb 2025) has recorded scores on 11 benchmarks, each shown with its evidence status.
Gemini 2.0 Flash (Feb 2025) has recorded scores on 11 benchmarks — GPQA diamond, MATH level 5, OTIS Mock AIME 2024-2025, DTBench, WeirdML, CadEval, and 5 more. The full table above shows each score with its evidence status.
4 of 11 recorded scores (36%) are independently reproduced rather than self-reported by the lab.
Gemini 2.0 Flash (Feb 2025) has 11 tracked claims: 4 independently reproduced, 7 unverified.