OpenAI's GPT-6 flagship for complex reasoning, coding, computer use, research, and end-to-end professional work.
Terminal-Bench 4.0 in the Artificial Analysis evaluation harness.
Humanity's Last Exam as evaluated by Artificial Analysis; the source configuration is preserved.
GDPval-AA v2.1 professional knowledge-work Elo. This is a task lens, not an assessment of every research skill.
GDP.pdf professional document understanding, including visual tables, charts and diagrams; not an audio/video score.
AutomationBench-AA multi-step business workflows in the Artificial Analysis harness.
Source-backed rows with the original evaluation context preserved.
| Benchmark | Score | Source | Evaluation notes |
|---|---|---|---|
| Artificial Analysis Coding Agent Index v1.4Evaluated 2026-09-03 | 67 | OpenAI | OpenAI comparison-table Artificial Analysis Coding Agent Index v1.4 result; score unit is index. |
| AutomationBench-AAChecked 2026-10-02 | 68% | Artificial Analysis | Artificial Analysis Intelligence Index v4.3.2; configuration=max. Rounded public score; checked date is not the evaluation execution date. |
| AA GDP.pdfChecked 2026-10-02 | 31% | Artificial Analysis | Artificial Analysis Intelligence Index v4.3.2; configuration=max. Rounded public score; checked date is not the evaluation execution date. |
| GDPval-AA v2.1Checked 2026-10-02 | 1542 | Artificial Analysis | Artificial Analysis Intelligence Index v4.3.2; configuration=max. Rounded public score; checked date is not the evaluation execution date. |
| AA Humanity's Last ExamChecked 2026-10-02 | 55% | Artificial Analysis | Artificial Analysis Intelligence Index v4.3.2; configuration=max. Rounded public score; checked date is not the evaluation execution date. |
| Artificial Analysis Intelligence Index v4.3.2Checked 2026-10-02 | 53 | Artificial Analysis | Artificial Analysis Intelligence Index v4.3.2; configuration=max. Rounded public score; checked date is not the evaluation execution date. |
| AA Terminal-Bench 4.0Checked 2026-10-02 | 59% | Artificial Analysis | Artificial Analysis Intelligence Index v4.3.2; configuration=max. Rounded public score; checked date is not the evaluation execution date. |
| Agents' Last ExamEvaluated 2026-09-03 | 59.3% | OpenAI | OpenAI-reported Agents' Last Exam result; highest-scoring setting. |
| ARC-AGI-3Evaluated 2026-09-03 | 99.9% | OpenAI | OpenAI-reported ARC-AGI-3 result; Responses API harness changes two settings. |
| Artificial Analysis Intelligence IndexEvaluated 2026-09-04 | 61 | Artificial Analysis | Artificial Analysis Intelligence Index v4.1.1; GPT-6 Astra max reasoning configuration; 9-evaluation composite. |
| Artificial Analysis Intelligence IndexEvaluated 2026-09-04 | 60 | Artificial Analysis | Artificial Analysis Intelligence Index v4.1.1; GPT-6 Astra high reasoning configuration. |
| Artificial Analysis Intelligence IndexEvaluated 2026-09-04 | 57 | Artificial Analysis | Artificial Analysis Intelligence Index v4.1.1; GPT-6 Astra low reasoning configuration. |
| BenchCADEvaluated 2026-09-03 | 95.9% | OpenAI | OpenAI-reported BenchCAD geometric-overlap score with tools; benchmark modifications are documented in the release notes. |
| DeepSWE v1.1Evaluated 2026-09-03 | 74.1% | OpenAI | OpenAI comparison-table result; preserve the source-specific agent harness. |
| ExploitBenchEvaluated 2026-09-03 | 100% | OpenAI | OpenAI-reported ExploitBench result without production safeguards; cybersecurity evaluation. |
| ExploitGymEvaluated 2026-09-03 | 42.4% | OpenAI | OpenAI-reported ExploitGym result without production safeguards; cybersecurity evaluation. |
| FrontierCode 1.1 ExtendedEvaluated 2026-09-03 | 64.5% | OpenAI | OpenAI-reported FrontierCode 1.1 Extended result; developer-message and harness notes apply. |
| FrontierCode 1.1 MainEvaluated 2026-09-03 | 53.3% | OpenAI | OpenAI-reported FrontierCode 1.1 Main result; developer-message and harness notes apply. |
| FrontierMath Tier 4 (v2)Evaluated 2026-09-03 | 97.6% | OpenAI | OpenAI-reported FrontierMath Tier 4 (v2) result. |
| GPQA / GPQA DiamondEvaluated 2026-09-03 | 96% | OpenAI | OpenAI-reported GPQA Diamond result. |
| Humanity's Last Exam with ToolsEvaluated 2026-09-03 | 57.2% | OpenAI | OpenAI-reported Humanity's Last Exam result with tools. |
| Database Migration Tasks (Internal)Evaluated 2026-09-03 | 63.9% | OpenAI | OpenAI internal Database Migration Tasks result. |
| Mind2WebEvaluated 2026-09-03 | 1.9 | OpenAI | OpenAI-reported Codex harness task-completion speedup versus the GPT-5.6 Sol experience. |
| OSWorld 2.0Evaluated 2026-09-03 | 72.6% | OpenAI | OpenAI latency-simulation result on OSWorld 2.0; approximately 40 minutes per task. This is not directly comparable to offline or modified-task rows. |
| SRE-BenchEvaluated 2026-09-03 | 88% | OpenAI | OpenAI-reported SRE-Bench single-attempt success rate. |
| SRE-BenchEvaluated 2026-09-03 | 99.2% | OpenAI | OpenAI-reported SRE-Bench success rate within four attempts; distinct evaluation setting. |
| Terminal-Bench 4.0Evaluated 2026-09-03 | 57.9% | OpenAI | OpenAI-reported Terminal-Bench 4.0 result. |
| Terminal-Bench-Science 0.1Evaluated 2026-09-03 | 64.6% | OpenAI | OpenAI-reported Terminal-Bench Science 0.1 result. |
Official OpenAI release report with GPT-6 Astra benchmark tables, computer-use and cybersecurity evaluations, and methodology footnotes.
Official API documentation for gpt-6-astra limits, modalities, reasoning levels, knowledge cutoff, supported tools, and token pricing.
Underlying raw metrics and exact model configurations remain authoritative.
Underlying raw metrics and exact model configurations remain authoritative.