GPT-5.6 Sol vs GPT-6 Astra
Humanity's Last Exam Humanity's Last Exam: Very hard expert questions across many fields, written to stump today’s models
GPT-5.6 Sol42.2
GPT-6 Astra52.7
Terminal-Bench 4.0 Terminal-Bench 4.0 · source setting 2: Agents use a terminal to complete executable coding, system administration and research tasks in the 4.0 suite
GPT-5.6 Sol14.6
GPT-6 Astra49.5
AutomationBench-AA AutomationBench-AA: Agents orchestrate SaaS workflows through APIs; objectives receive credit only when the task has no guardrail violation
GPT-5.6 Sol51.3
GPT-6 Astra64.6
Arena Text · creative writing Arena Text · creative writing: Human pairwise preferences in the stated text vote pool
GPT-5.6 Sol · xhigh1,469
GPT-6 Astra · max1,450
MMMU-Pro (vision) MMMU-Pro (vision): College-level questions that require understanding images: charts, diagrams, photos
GPT-5.6 Sol81.4
GPT-6 Astra85.1
| Benchmark | GPT-5.6 Sol | GPT-6 Astra |
|---|---|---|
| Humanity's Last Exam | 42.2 | 52.7 |
| Terminal-Bench 4.0 | 14.6 | 49.5 |
| AutomationBench-AA | 51.3 | 64.6 |
| Arena Text · creative writing | 1,469 · xhigh | 1,450 · max |
| MMMU-Pro (vision) | 81.4 | 85.1 |
Specs
| Spec | GPT-5.6 Sol | GPT-6 Astra |
|---|---|---|
| Input, per 1M | $4 | $10 |
| Output, per 1M | $20 | $50 |
| Cost per task? | $0.50 | $1.54 |
| Speed, tok/s | 85 | 59 |
| Context | 1.05M | 1.05M |
| Weights | Closed | Closed |
| Released | Jul 2026 | Sep 2026 |
| See images | ✓ | ✓ |
| Read PDFs | Partial | Partial |
| Hear audio | — | — |
| Watch video | — | — |
| Call tools | ✓ | ✓ |
| Force a tool call | Partial | Partial |
| Return strict JSON | ✓ | ✓ |
| Search the web | ? | ✓ |
| Run code | ? | ✓ |
| Use a computer | ? | ✓ |
What breaks when you switch
GPT-5.6 Sol → GPT-6 Astra
Monitoring can return HTTP 403›
Misalignment monitoring can return HTTP 403 before streaming or errors after output; match misalignment_policy_violation and handle stream errors.
No none reasoning effort›
The none reasoning effort is unsupported on GPT-6 Astra and GPT-6.1 Sol; use low instead.
Remove unsupported sampling parameters›
When reasoning effort is not none, remove temperature, top_p and top_logprobs, plus Chat Completions logprobs or Responses output logprobs inclusion.
GPT-6 Astra → GPT-5.6 Sol
Nothing known.