GPT-6 Astra (xHigh): benchmark results
Provider: OpenAI. Released 2026-09-03. Access: API.
Aggregate Rating 1968 ± 8, rank #12 of 2257 rated models, from 284 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| AA GPQA Diamond | 96.26 | Accuracy (%) | 100 |
| DeepSWE | 74.1 | Pass@1 (%; DeepSWE v1.1 scored attempts) | 100 |
| DeepSWE-mini (16 Tasks) | 81.25 | Pass@1 (%, 16-task subset) | 100 |
| Epoch AI - Weirdml V3 | 42.22 | Score | 100 |
| FrontierCS 2 - Go1 Joystick Policy Training | 50.56 | Task score (0–100) | 100 |
| FrontierCS 2 - Latent Objective Recovery | 99.49 | Task score (0–100) | 100 |
| FrontierCS 2 - SQL Counterexample Search | 97.22 | Task score (0–100) | 100 |
| FrontierCS 2 - Sparse Video Attention Policy | 48.59 | Task score (0–100) | 100 |
| LLM2014 Logic 2026-09 | 88.75 | Median Score | 100 |
| LLM2014 Logic 2026-10 | 86.53 | Median Score | 100 |
| LLM2014 Logic 2026-11 | 86.35 | Median Score | 100 |
| NonoBench | 100 | Standard accuracy (%), correct / non-failed attempts; versio | 100 |
Interactive version: theaggregate.ai/model?slug=gpt-6-astra-xhigh · How It Works · Data refreshed daily, snapshot 2026-11-17.