text-davinci-002 — benchmark results
InstructGPT completion model snapshot from the text-davinci series. Provider: OpenAI. Released 2022-03-15. Access: API.
Unified ELO 1364 ± 18, rank #1362 of 1776 rated models, from 39 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| HELM Classic - BBQ | 88.07 | Exact Match (%) | 100 |
| HELM Classic - TruthfulQA | 60.96 | Exact Match (%) | 98.5 |
| HELM Classic - WikiFact | 39.25 | Exact Match (%) | 97 |
| HELM Classic - HellaSwag | 81.5 | Exact Match (%) | 96.8 |
| HELM Classic - OpenbookQA | 59.4 | Exact Match (%) | 96.8 |
| HELM Classic - Synthetic Reasoning Natural | 62.29 | F1 (%) | 95.6 |
| HELM Classic - BoolQ | 87.7 | Exact Match (%) | 95.5 |
| HELM Classic - CivilComments | 66.84 | Exact Match (%) | 95.5 |
| HELM Classic - Entity Data Imputation | 84.16 | Exact Match (%) | 95.5 |
| HELM Classic - Entity Matching | 93.08 | Exact Match (%) | 95.5 |
| HELM Classic - NaturalQuestions Open Book | 71.32 | F1 (%) | 95.4 |
| HELM Classic - MATH | 32.79 | Equivalent (%) | 94.1 |
Interactive version: theaggregate.ai/model?slug=text-davinci-002 · How the rankings work · Data refreshed daily, snapshot 2026-07-22.