Claude Opus 5 vs GPT-6 Astra
Arena Text · creative writing Arena Text · creative writing: Human pairwise preferences in the stated text vote pool
Opus 5 · max1,471
GPT-6 Astra · max1,450
EQ-Bench Longform Creative Writing 1.11 Longform Creative Writing 1.11: Models plan, revise and write a story across eight turns, graded with a writing rubric
Opus 586.3
GPT-6 Astra82.8
EQ-Bench Creative Writing v3 Creative Writing v3 · Elo: Judge-rated creative writing under the v3 prompt set and pairwise ranking
Opus 52,132.6
GPT-6 Astra2,173.3
Lech Mazur Short-Story Lech Mazur Short-Story · comparison score: Judges compare stories written to ten required elements; the comparison score can be negative
Opus 5 · xhigh3.7
GPT-6 Astra · high3.4
Harvey LAB-AA 1.1 Harvey LAB-AA 1.1: Legal agents work from case documents; every rubric criterion must pass and material hallucinations receive no credit
Opus 5—
GPT-6 Astra · max8.6
| Benchmark | Opus 5 | GPT-6 Astra |
|---|---|---|
| Arena Text · creative writing | 1,471 · max | 1,450 · max |
| EQ-Bench Longform Creative Writing 1.11 | 86.3 | 82.8 |
| EQ-Bench Creative Writing v3 | 2,132.6 | 2,173.3 |
| Lech Mazur Short-Story | 3.7 · xhigh | 3.4 · high |
| Harvey LAB-AA 1.1 | — | 8.6 · max |
Specs
| Spec | Opus 5 | GPT-6 Astra |
|---|---|---|
| Input, per 1M | $5 | $10 |
| Output, per 1M | $25 | $50 |
| Cost per task? | $2.19 | $1.54 |
| Speed, tok/s | 55 | 59 |
| Context | 1M | 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
Opus 5 → 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 → Opus 5