Updated daily at 05:00 UTC
- Released
- 14 Mar 2023
- Status
- Deprecated
- Retired
- 23 Oct 2026
- Context
- 8.2K
- Max output
- 8.2K
- Input per 1M tokens
- $30.00
- Output per 1M tokens
- $60.00
- Open weights
- No
- License
- proprietary
- Parameters
- —
- Knowledge cutoff
- —
- Reasoning
- No
- Modalities
- text → text
Category scores
Out of 100, from independent and third-party results only, by method category-2026-09-29.1.
| Category | Score | Rank | Coverage |
|---|---|---|---|
| Knowledge & reasoning | 10.0 | 95 | 2 of 4 |
| Human preference | 34.4 | Not ranked | 1 of 5 |
Benchmark results
Every result we hold for this model, with who ran it and where it was published.
Knowledge & reasoning
| Benchmark | Metric | Value | Provenance | Setting | Evaluated | Source |
|---|---|---|---|---|---|---|
| BIG-Bench Hard (BBH) | Average accuracy | 75.1% | Third party | — | — | Baichuan 2: Open Large-scale Language Models via Epoch AI |
| Chess Puzzles | Puzzles solved | 4.0% | Independent | — | 7 Aug 2026 | Epoch AI |
| DTBench (Decision Theory Benchmark) | Accuracy (capability questions) | 62.7% | Third party | — | — | https://conceptualreasoning.ai/dtbench via Epoch AI |
| HellaSwag | Accuracy | 95.3% | Self-reported | — | — | GPT-4 Technical Report via Epoch AI |
| LMCA (Language Model Conceptual Argumentation) | Correlation with expert ratings (Pearson r) | 0.1706 | Third party | — | — | https://conceptualreasoning.ai/lmca via Epoch AI |
| MMLU | Accuracy | 82.4% | Third party | — | — | Stanford CRFM Leaderboard via Epoch AI |
| MMLU | Accuracy | 86.4% | Self-reported | — | — | GPT-4 Technical Report via Epoch AI |
| Mystery Game Puzzles | Puzzles solved | 12.0% | Independent | — | 28 Aug 2026 | Epoch AI |
| WinoGrande | Accuracy | 87.5% | Self-reported | — | — | GPT-4 technical report via Epoch AI |
Scientific reasoning
| Benchmark | Metric | Value | Provenance | Setting | Evaluated | Source |
|---|---|---|---|---|---|---|
| GPQA Diamond | Accuracy | 35.7% | Independent | — | 23 Oct 2025 | Epoch AI |
| GPQA Diamond | Accuracy | 30.7% | Independent | — | 27 Jan 2025 | Epoch AI |
Mathematics
| Benchmark | Metric | Value | Provenance | Setting | Evaluated | Source |
|---|---|---|---|---|---|---|
| GSM8K | Exact match | 90.0% | Third party | — | — | Baichuan 2: Open Large-scale Language Models via Epoch AI |
| GSM8K | Exact match | 92.0% | Self-reported | — | — | GPT-4 Technical Report via Epoch AI |
| MATH Level 5 | Accuracy | 23.0% | Independent | — | 27 Jan 2025 | Epoch AI |
| OTIS Mock AIME 2024-2025 | Accuracy | 1.1% | Independent | — | 23 Oct 2025 | Epoch AI |
| OTIS Mock AIME 2024-2025 | Accuracy | 0.6% | Independent | — | 22 Oct 2025 | Epoch AI |
Coding & software engineering
| Benchmark | Metric | Value | Provenance | Setting | Evaluated | Source |
|---|---|---|---|---|---|---|
| METR Time Horizons | Average task success (Epoch's column) | 36.1% | Third party | — | — | METR - Measuring AI Ability to Complete Long Tasks via Epoch AI |
| METR Time Horizons | Average task success (Epoch's column) | 29.3% | Source not stated | — | — | Epoch AI via Epoch AI |
| WeirdML (v2) | Average accuracy | 12.4% | Third party | — | — | https://htihle.github.io/weirdml.html via Epoch AI |
Factuality & hallucination
| Benchmark | Metric | Value | Provenance | Setting | Evaluated | Source |
|---|---|---|---|---|---|---|
| TriviaQA | Exact match | 84.8% | Third party | — | — | PapersWithCode via Epoch AI |
Human preference
| Benchmark | Metric | Value | Provenance | Setting | Evaluated | Source |
|---|---|---|---|---|---|---|
| Text Arena (arena.ai) | Arena score (Elo scale) | 1288 | Independent | — | 25 Sep 2026 | Arena |
| Text Arena (arena.ai) | Arena score (Elo scale) | 1276 | Independent | — | 25 Sep 2026 | Arena |
| Text Arena (arena.ai) | Votes | 88723 | Independent | — | 25 Sep 2026 | Arena |
| Text Arena (arena.ai) | Votes | 54173 | Independent | — | 25 Sep 2026 | Arena |
Other GPT-4 versions
- GPT-4 Turbo6 Nov 2023
- GPT-4 32K14 Mar 2023