GPT-6 Astra: benchmark results
Can continue working while tools run and accept corrections mid-task. Reasoning stays enabled even at its lowest effort setting. Provider: OpenAI. Released 2026-09-03. Access: API.
Aggregate Rating 1949 ± 4, rank #8 of 10556 rated models, from 2161 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| 3DVerifier - Explanation-aware Overall | 65.8 | Explanation-aware balanced accuracy (%) | 100 |
| 3DVerifier - Material | 71.5 | Balanced accuracy (%) | 100 |
| 3DVerifier - Motion Binary | 57.5 | Balanced accuracy (%) | 100 |
| 3DVerifier - Motion Four-class | 54.5 | Four-class accuracy (%) | 100 |
| 3DVerifier - Overall | 69.4 | Balanced accuracy (%) | 100 |
| 3DVerifier - Part | 72.1 | Balanced accuracy (%) | 100 |
| 3DVerifier - Part Grounding | 89.6 | Balanced accuracy (%) | 100 |
| 3DVerifier - Part Relation | 90.3 | Balanced accuracy (%) | 100 |
| 4DCodeBench | 0.79 | Reconstruction score (0–1), five-family mean over 200 scenes | 100 |
| AA GPQA Diamond | 96.26 | Accuracy (%) | 100 |
| AI Coding Daily (Codex CLI) - Laravel Code Quality | 18.6 | Laravel Code Quality (max 20) points, LLM-judged rubric scor | 100 |
| AI Coding Daily (Codex CLI) - Python Code Quality | 18.5 | Python Code Quality (max 20) points, LLM-judged rubric score | 100 |
Interactive version: theaggregate.ai/model?slug=gpt-6-astra · How It Works · Data refreshed daily, snapshot 2026-11-17.