OpenAI·released 2024-04-091 source
GPT-4 Turbo (Apr 2024) benchmark scores: 9 benchmarks tracked. 44% of its scores are independently reproduced — the rest are self-reported or unverified.
Leads on 0 of 9 benchmarks·best result 75.07 on MMLU (unverified)·4 of 9 independently reproduced
| Benchmark | Score | Evidence status | Measured |
|---|---|---|---|
| MMLU | 75.07accuracy (%) | unverified· optimizedT1 | 2024-04-09Stanford CRFM Leaderboard |
| MATH level 5 | 46.73accuracy (%) | reproduced· optimizedT1 | 2024-04-09Epoch AI |
| METR Time Horizons | 36.74accuracy (%) | unverifiedT1 | 2024-04-09METR - Measuring AI Ability to Complete Long Tasks |
| GPQA diamond | 28.79accuracy (%) | reproduced· optimizedT1 | 2024-04-09Epoch AI |
| WeirdML | 18.01accuracy (%) | unverifiedT1 | 2024-04-09https://htihle.github.io/weirdml.html |
| SimpleBench | 10.12accuracy (%) | unverifiedT1 | 2024-04-09SimpleBench Leaderboard |
| OTIS Mock AIME 2024-2025 | 6.57accuracy (%) | reproduced· optimizedT1 | 2024-04-09Epoch AI |
| Chess Puzzles | 1.09accuracy (%) | reproducedT1 | 2024-04-09Epoch AI |
| FrontierMath-Tiers-1-3-v2-Private | 0.7accuracy (%) | unverifiedT2 | 2024-04-09 |
Claims drawn from cited facts, not live model generation.
This model comes from OpenAI and was released in 2024, placing it within the developer's recent generation of systems. That vintage means it reflects contemporary training and alignment practice rather than an earlier era of design, so its provenance and timing are the main signals of what to expect from it.
4 cited facts
This model is scored on 9 tracked benchmarks, but it currently holds no top score on any of them. Its record can be trusted: the results have been independently reproduced, meaning outside evaluation backs the scores up.
3 cited facts
Under the disclosed harness, it trails the state-of-the-art average by 63.01 points, a wide gap that places it well behind the leaders. It holds the top score on none of its benchmarks, indicating no category leadership yet.
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
GPT-4 Turbo (Apr 2024) is an AI model developed by OpenAI, released 2024-04-09. Its benchmark record below tags every score as independently reproduced, self-reported, or unverified.
GPT-4 Turbo (Apr 2024) has recorded scores on 9 benchmarks, each shown with its evidence status.
GPT-4 Turbo (Apr 2024) has recorded scores on 9 benchmarks — GPQA diamond, MATH level 5, OTIS Mock AIME 2024-2025, WeirdML, MMLU, Chess Puzzles, and 3 more. The full table above shows each score with its evidence status.
4 of 9 recorded scores (44%) are independently reproduced rather than self-reported by the lab.
GPT-4 Turbo (Apr 2024) has 9 tracked claims: 4 independently reproduced, 5 unverified.