GPT-3.5 Turbo (0613): benchmark results
OpenAI's GPT-3.5 Turbo (0613), a June 2023 update adding function calling and more reliable system-message steerability, with a 4,096-token context window. Provider: OpenAI. Released 2023-06-13. Access: API.
Aggregate Rating 1427 ± 10, rank #1118 of 1656 rated models, from 556 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| HELM CLEVA - Cleva Mathematical Calculation | 99.16 | EM | 100 |
| HELM CLEVA - Cleva Translation | 45.25 | BLEU | 100 |
| HELM Classic - CivilComments | 69.64 | Exact Match (%) | 100 |
| HELM Classic - MATH Chain-of-Thought | 71.86 | Equivalent (%) | 100 |
| HELM Classic - MATH | 45.28 | Equivalent (%) | 98.5 |
| HELM Classic - Synthetic Reasoning Abstract | 50.94 | Exact Match (%) | 98.5 |
| FastEval | 65.17 | Total Score | 96.9 |
| HELM Classic - QuAC | 48.5 | F1 (%) | 96.9 |
| HELM Classic - RAFT | 74.77 | Exact Match (%) | 95.5 |
| HELM AIR-Bench 2024 - #44.3: Sexual exploitation | 100 | Refusal Rate (%) | 94.2 |
| HELM Classic - Synthetic Reasoning Natural | 58.56 | F1 (%) | 94.1 |
| HELM AIR-Bench 2024 - #6.24: Accounting | 73.33 | Refusal Rate (%) | 93 |
Interactive version: theaggregate.ai/model?slug=gpt-3-5-turbo-0613 · How It Works · Data refreshed daily, snapshot 2026-10-20.