Google DeepMind·released 2026-05-191 source
Gemini 3.5 Flash benchmark scores: 19 benchmarks tracked. 37% of its scores are independently reproduced — the rest are self-reported or unverified.
Leads on 0 of 19 benchmarks·best result 95.55 on OTIS Mock AIME 2024-2025 (reproduced)·7 of 19 independently reproduced·$1.5/$9 per M tokens
Consensus: LiteLLM · Cross-check: models.dev, OpenRouter · See every model’s pricing →
| Benchmark | Score | Evidence status | Measured |
|---|---|---|---|
| OTIS Mock AIME 2024-2025 | 95.55accuracy (%) | reproduced· optimizedT1 | 2026-05-19Epoch AI |
| ARC-AGI | 92.5accuracy (%) | unverified· optimizedT1 | 2026-05-19https://arcprize.org/leaderboard |
| DTBench | 91.12accuracy (%) | unverifiedT1 | 2026-05-19https://conceptualreasoning.ai/dtbench |
| GPQA diamond | 90.4accuracy (%) | reproduced· optimizedT1 | 2026-05-19Epoch AI |
| SWE-Bench verified | 79.34accuracy (%) | reproduced· optimizedT1 | 2026-05-19Epoch AI |
| ARC-AGI-2 | 72.08accuracy (%) | unverifiedT2 | 2026-05-19 |
| SimpleBench | 72.04accuracy (%) | unverifiedT1 | 2026-05-19SimpleBench Leaderboard |
| SimpleQA Verified | 66.2accuracy (%) | unverifiedT2 | 2026-05-19 |
| FrontierMath-Tiers-1-3-v2-Private | 62.81accuracy (%) | reproducedT1 | 2026-05-19Epoch AI |
| WeirdML | 62.64accuracy (%) | unverifiedT1 | 2026-05-19https://htihle.github.io/weirdml.html |
| Surface Evolver Bench | 58.13accuracy (%) | unverifiedT1 | 2026-05-19https://yhenon.github.io/surface-evolver-llm-eval/ |
| LMCA | 55.47accuracy (%) | unverifiedT1 | 2026-05-19https://conceptualreasoning.ai/lmca |
| Chess Puzzles | 42.13accuracy (%) | unverifiedT2 | 2026-05-19 |
| DeepSWE | 37.39accuracy (%) | unverifiedT1 | 2026-05-19https://deepswe.datacurve.ai/ |
| ProofBench | 31accuracy (%) | unverifiedT1 | 2026-05-19https://www.vals.ai/benchmarks/proof_bench |
| APEX-Agents | 27.5accuracy (%) | unverifiedT2 | 2026-05-19 |
| FrontierMath-Tier-4-v2-Private | 26.83accuracy (%) | reproducedT1 | 2026-05-19Epoch AI |
| Mystery Game Puzzles | 25.09accuracy (%) | reproducedT1 | 2026-05-19Epoch AI |
| EBR-bench | 4.76accuracy (%) | reproducedT1 | 2026-05-19Epoch AI |
Claims drawn from cited facts, not live model generation.
This model was developed by Google DeepMind. It was released in 2026.
2 cited facts
This model is scored on 19 tracked benchmarks. It currently holds no top score among them. The record’s evidence status is independently reproduced.
3 cited facts
Its mean gap to the best recorded score across its tracked benchmarks is 30.63 points. It currently holds the top score on none of them.
2 cited facts
6 facts cross-checked across data sources: 3 corroborated, 2 single-source, 1 disputed.
Source facts, citations, and refresh stamp for this record.
Sources: epoch_benchmarks, litellm_prices, modelsdev_models, openrouter_models
Gemini 3.5 Flash is an AI model developed by Google DeepMind, released 2026-05-19. Its benchmark record below tags every score as independently reproduced, self-reported, or unverified.
Gemini 3.5 Flash has recorded scores on 19 benchmarks, each shown with its evidence status.
Gemini 3.5 Flash has recorded scores on 19 benchmarks — GPQA diamond, OTIS Mock AIME 2024-2025, DTBench, WeirdML, LMCA, SimpleQA Verified, and 13 more. The full table above shows each score with its evidence status.
7 of 19 recorded scores (37%) are independently reproduced rather than self-reported by the lab.
Gemini 3.5 Flash has 19 tracked claims: 7 independently reproduced, 12 unverified.
Listed API pricing: $1.5 per million input tokens, $9 per million output tokens. See the pricing block for the full breakdown.