DeepSeek·released 2026-09-091 source
DeepSeek V4.1 Flash benchmark scores: 16 benchmarks tracked. 0% of its scores are independently reproduced — the rest are self-reported or unverified.
Leads on 0 of 16 benchmarks·best result 98.33 on OTIS Mock AIME 2024-2025 (unverified)·0 of 16 independently reproduced
| Benchmark | Score | Evidence status | Measured |
|---|---|---|---|
| OTIS Mock AIME 2024-2025 | 98.33accuracy (%) | unverifiedT2 | 2026-09-09 |
| ARC-AGI | 94.5accuracy (%) | unverifiedT1 | 2026-09-09https://arcprize.org/leaderboard |
| GPQA diamond | 86.45accuracy (%) | unverifiedT2 | 2026-09-09 |
| DTBench | 83.12accuracy (%) | unverifiedT1 | 2026-09-09https://conceptualreasoning.ai/dtbench |
| ARC-AGI-2 | 72.92accuracy (%) | unverifiedT2 | 2026-09-09 |
| FrontierMath-Tiers-1-3-v2-Private | 67.37accuracy (%) | unverifiedT2 | 2026-09-09 |
| LMCA | 55.26accuracy (%) | unverifiedT1 | 2026-09-09https://conceptualreasoning.ai/lmca |
| ProofBench | 54accuracy (%) | unverifiedT1 | 2026-09-09https://www.vals.ai/benchmarks/proof_bench |
| Surface Evolver Bench | 46.25accuracy (%) | unverifiedT1 | 2026-09-09https://yhenon.github.io/surface-evolver-llm-eval/ |
| Chess Puzzles | 45.29accuracy (%) | unverifiedT2 | 2026-09-09 |
| FrontierCode | 43.87accuracy (%) | unverifiedT1 | 2026-09-09https://cognition.com/frontiercode |
| SimpleQA Verified | 42.9accuracy (%) | unverifiedT2 | 2026-09-09 |
| APEX-Agents | 39.5accuracy (%) | unverifiedT2 | 2026-09-09 |
| Mystery Game Puzzles | 37.21accuracy (%) | unverifiedT2 | 2026-09-09 |
| FrontierMath-Tier-4-v2-Private | 26.83accuracy (%) | unverifiedT2 | 2026-09-09 |
| Furniture Assembly | 5.95accuracy (%) | unverifiedT2 | 2026-09-09 |
Claims drawn from cited facts, not live model generation.
This model was developed by DeepSeek and released in 2026.
2 cited facts
This model has scores on 16 tracked benchmarks and currently leads 0 of them. The record’s evidence status is unverified.
3 cited facts
Its mean gap to the best recorded score across its tracked benchmarks is 31.27 points.
1 cited fact
1 facts cross-checked across data sources: 1 single-source.
Source facts, citations, and refresh stamp for this record.
Sources: epoch_benchmarks
DeepSeek V4.1 Flash is an AI model developed by DeepSeek, released 2026-09-09. Its benchmark record below tags every score as independently reproduced, self-reported, or unverified.
DeepSeek V4.1 Flash has recorded scores on 16 benchmarks, each shown with its evidence status.
DeepSeek V4.1 Flash has recorded scores on 16 benchmarks — GPQA diamond, OTIS Mock AIME 2024-2025, DTBench, LMCA, SimpleQA Verified, Chess Puzzles, and 10 more. The full table above shows each score with its evidence status.
0 of 16 recorded scores (0%) are independently reproduced rather than self-reported by the lab.
DeepSeek V4.1 Flash has 16 tracked claims: 16 unverified.