GPT-4 Turbo (Jan 2024)
GPT-4 Turbo (Jan 2024) ranks first on BABILong.
18
Benchmarks with a score
1
Ranked first, out of five or more models
2
In the top three
OpenAI Jan 2024
GPT-4 Turbo (Jan 2024)
Closed weights Flagship
No Epoch Capabilities Index score yet
- MathBench 58.8% #5 /21
- CFBench 64.7% #4 /19
- Situational Awareness Dataset (SAD) 51.5% #5 /17
Where it ranks
Its best score on each benchmark, and where that ranks among every model with a score there. We rank each model once, at its best setting.
Knowledge and language
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| EQ-Bench v2 | 83.9 | #7 of 85 | Official leaderboard | |
| CulturalBench | 61.6% | #2 of 15 | Official leaderboard | |
| WildBench | 53.3 | #7 of 47 | Official leaderboard | |
| BABILong | 72.4% | #1 of 6 | Official leaderboard | |
| CFBench | 64.7% | #4 of 19 | Official leaderboard | |
| Arena-Hard-Auto | 78% | #10 of 38 | Official leaderboard | |
| RewardBench | 84.3% | #17 of 48 | Official leaderboard | |
| LMArena (Elo) | 1,313 | #218 of 335 | Official leaderboard |
Math and science
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| MathBench | 58.8% | #5 of 21 | Official leaderboard |
Coding
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| DevBench Implementation | 7.10% | #1 of 3 | Official leaderboard | |
| Aider-Edit | 66.2% | #18 of 34 | Official leaderboard | |
| METR Time Horizon | 6.3 min | #25 of 31 | Official leaderboard |
Safety and security
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| OR-Bench | 12.2% | #4 of 28 | Official leaderboard | |
| Situational Awareness Dataset (SAD) | 51.5% | #5 of 17 | Official leaderboard | |
| SORRY-Bench | 26.4% | #10 of 23 | Official leaderboard | |
| NYU CTF Bench | 5.00% | #5 of 7 | Official leaderboard | |
| AgentDojo | 56.3% | #20 of 20 | Official leaderboard | |
| JailbreakBench | 0% | #1 of 1 | Official leaderboard |