GPT-6 Astra
GPT-6 Astra is OpenAI's most capable model for complex reasoning, coding, computer use, research, and document creation.
Shogo Agent Index
Supported#4 of 9270.9range 66.5–75.4 · 38% of weight measured
Agentic coding
80
43% measured
Long-horizon autonomy
80
50% measured
Knowledge-work agents
56
52% measured
Computer use
67
estimated
Security & context learning
63
estimated
Agentic benchmarks
| Benchmark | Score | Normalised | Index weight | Harness / source |
|---|---|---|---|---|
| SWE-bench Verified | est. | 8% | predicted 90 | |
| Terminal-Bench 2.0 | est. | 8% | predicted 97 | |
| DeepSWE | 74.1% | 6% | mini-swe-agent · $4.43/task | |
| FrontierSWE | 65.5% | 5% | proximus · $1029.65/task | |
| CursorBench | est. | 4% | predicted 56 | |
| GSO-Bench | 79.4% | 4% | OpenHands | |
| METR Time Horizons | est. | 10% | predicted 87 | |
| Vending-Bench 2 | $15,515 | 6% | source | |
| PostTrainBench | 41.9% | 4% | Codex CLI | |
| APEX-Agents | 64.7% | 8% | source | |
| GDPval | est. | 7% | predicted 72 | |
| Remote Labor Index | 20.8% | 5% | source | |
| DeepResearch Bench | est. | 5% | predicted 53 | |
| OSWorld 2.0 | est. | 6% | predicted 47 | |
| OSWorld | est. | 4% | predicted 96 | |
| Cybench | est. | 5% | predicted 95 | |
| CL-bench | est. | 5% | predicted 31 |
Other benchmarks
| Benchmark | Category | Score | Source |
|---|---|---|---|
| ARC-AGI | Reasoning | 98.5% | source |
| ARC-AGI-2 | Reasoning | 95.0% | source |
| BALROG | Long-horizon | 68.3% | source |
| Chess Puzzles | Reasoning | 76.0% | source |
| CritPt | Reasoning | 31.7% | source |
| DTBench | Knowledge | 97.3% | source |
| FrontierCode | Coding | 53.3% | codex |
| FrontierMath Tiers 1-3 | Math | 93.7% | source |
| FrontierMath Tier 4 | Math | 97.6% | source |
| GPQA Diamond | Reasoning | 95.8% | source |
| Humanity's Last Exam | Reasoning | 54.8% | source |
| LMCA | Knowledge | 64.4% | source |
| MirrorCode | Coding | 60.0% | source |
| OTIS Mock AIME | Math | 100.0% | source |
| ProofBench | Math | 99.0% | source |
| SciCode | Coding | 56.5% | source |
| SimpleBench | Reasoning | 86.5% | source |
| SimpleQA Verified | Knowledge | 75.6% | source |
| WebDev Arena | Coding | 1800 | source |
| WeirdML v3 | Coding | 42.2% | codex_cli |
API hosts
Live prices and uptime for every host serving GPT-6 Astra via OpenRouter.