Updated daily at 05:00 UTC
- Released
- 15 Mar 2022
- Status
- Retired
- Retired
- 4 Jan 2024
- Context
- —
- Max output
- —
- Input per 1M tokens
- $20.00
- Output per 1M tokens
- $20.00
- Open weights
- No
- License
- proprietary
- Parameters
- —
- Knowledge cutoff
- Jun 2021
- Reasoning
- No
- Modalities
- text → text
Category scores
Out of 100, from independent and third-party results only, by method category-2026-09-29.1.
Not enough independent results yet
Only independent and third-party results rank a model; self-reported and unattributed numbers never do.
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 |
|---|---|---|---|---|---|---|
| AI2 Reasoning Challenge (ARC) | Challenge set accuracy | 85.2% | Self-reported | — | — | GPT-4 Technical Report via Epoch AI |
| HellaSwag | Accuracy | 81.5% | Third party | — | — | Stanford HELM via Epoch AI |
| MMLU | Accuracy | 70.0% | Self-reported | — | — | GPT-4 Technical Report via Epoch AI |
| WinoGrande | Accuracy | 81.6% | Self-reported | — | — | GPT-4 technical report via Epoch AI |
Mathematics
| Benchmark | Metric | Value | Provenance | Setting | Evaluated | Source |
|---|---|---|---|---|---|---|
| GSM8K | Exact match | 41.5% | Third party | — | — | Stanford HELM via Epoch AI |
Other GPT-3.5 versions
- GPT-3.5 Turbo Instruct18 Sep 2023
- GPT-3.5 Turbo 16k13 Jun 2023
- GPT-3.5 Turbo1 Mar 2023
- text-davinci-00328 Nov 2022