GPT-3.5 Turbo (1106): benchmark results
November 2023 GPT-3.5 Turbo API snapshot (gpt-3.5-turbo-1106). Provider: OpenAI. Released 2023-11-06. Access: API.
Aggregate Rating 1456 ± 12, rank #961 of 1656 rated models, from 503 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| CodeEditorBench Plus - Code Translate (Zero-shot) | 47.5 | pass@1 (%, greedy decoding, zero-shot, CodeEditorBench_Plus) | 100 |
| HELM AIR-Bench 2024 - #6.24: Accounting | 76.67 | Refusal Rate (%) | 97.1 |
| Epoch AI - Adversarial Nli | 58.1 | Score | 95 |
| T-Eval | 84 | Overall Score (%) | 95 |
| HELM AIR-Bench 2024 - #6: Advice in Heavily Regulated Industries | 26 | Refusal Rate (%) | 94.8 |
| CodeEditorBench Plus - Code Requirement Switch (Zero-shot) | 17.7 | pass@1 (%, greedy decoding, zero-shot, CodeEditorBench_Plus) | 94.4 |
| ANLI | 58.1 | Score (%) | 93.8 |
| ARC Challenge (AI2) | 87.4 | Accuracy (%) | 92.1 |
| SpeechMap Compliance | 89.5 | % Requests Completed | 90.9 |
| OpenBookQA | 86 | Accuracy (%) | 90.5 |
| LogicKor - Grammar | 7.92 | Score (0-10) | 89.9 |
| HELM AIR-Bench 2024 - #6.22: Legal | 10 | Refusal Rate (%) | 89.5 |
Interactive version: theaggregate.ai/model?slug=gpt-3-5-turbo-1106 · How It Works · Data refreshed daily, snapshot 2026-10-20.