Claude Opus 5 vs Claude Sonnet 5.5
Vals Science Index?
Opus 547.9
Sonnet 5.559.1
| Benchmark | Opus 5 | Sonnet 5.5 |
|---|---|---|
| Humanity's Last Exam | 51.3 | 45.8 |
| Arena Text · overall | 1,490 · max | 1,475 · xhigh |
| Arena Text · Chinese | 1,573.1 · max | 1,534.6 · xhigh |
| Arena Text · English | 1,508.1 · max | 1,485.5 · xhigh |
| Arena Text · French | 1,518.4 · high | 1,494.6 · xhigh |
| Arena Text · instruction following | 1,518.4 · max | 1,490.5 · xhigh |
| Arena Text · Japanese | 1,520.5 · max | 1,459.9 · xhigh |
| Arena Text · non-English | 1,501.3 · max | 1,455.5 · xhigh |
| Arena Text · Russian | 1,507.9 · max | 1,458.1 · xhigh |
| Arena Text · Spanish | 1,522 · max | 1,446 · xhigh |
| AA Omniscience Index | 31 | 21 |
| SimpleQA Verified v1 | 63.3 | 46.5 · max |
| CritPt (physics) | 26.9 | 24.6 |
| AA Intelligence Index 4.3.2 | 44.8 | 46.8 |
| AA Omniscience · Accuracy | 57.1 | 52 |
| AA Omniscience · Hallucination Rate | 60.7 | 64.6 |
| AA-LCR (long context) 1.1 | 82 | 78 |
| GDP.pdf · AA all-pass | 20 | 25.2 |
| MLCR-AA | 13.9 | 18.9 |
| ARC-AGI-1 · public evaluation | 99 · max | 92.4 |
| ARC-AGI-1 · semi-private | 97.5 · max | 92 |
| ARC-AGI-2 · public evaluation | 97.1 · max | 70.7 |
| ARC-AGI-2 · semi-private | 90.4 · max | 67.4 |
| ARC-AGI-3 · semi-private | 30.2 · high | 1.2 |
| Chess Puzzles | 43 · max | 59 · max |
| FrontierMath Erdos | 0 · max | 2.9 · max |
| FrontierMath Tier 4 (v2) | 73.2 · max | 80.5 · max |
| FrontierMath Tiers 1–3 (v2) | 85.6 · max | 88.8 · max |
| Mystery Game Puzzles | 59 · max | 65 · max |
| Vals MedCode | 63.6 | 52.9 |
| Vals MedScribe | 91 | 91.1 |
| Vals Legal Research Bench | 55.3 | 48.1 |
| Vals Science Index | 47.9 | 59.1 |
| Arena Text · German | 1,530.1 · max | — |
| Arena Text · Korean | 1,518.1 · max | — |
| Arena Text · Polish | 1,522.9 · max | — |
| MathArena · ArXivMath · June 2026 | 82.6 · max | — |
| MathArena · BrokenArXiv · June 2026 | 90.7 · max | — |
| MathArena · ArXivLean · June 2026 | 32.6 · max | — |
| Vals LegalBench | 87 | — |
Specs
| Spec | Opus 5 | Sonnet 5.5 |
|---|---|---|
| Input, per 1M | $5 | $2 |
| Output, per 1M | $25 | $10 |
| Cost per task? | $2.19 | $0.88 |
| Speed, tok/s | 51 | 106 |
| Context | 1M | 1M |
| Weights | Closed | Closed |
| Released | Jul 2026 | Sep 2026 |
| See images | ✓ | ✓ |
| Read PDFs | Partial | ✓ |
| Hear audio | — | — |
| Watch video | — | — |
| Call tools | ✓ | ✓ |
| Force a tool call | Partial | — |
| Return strict JSON | ✓ | ✓ |
| Search the web | ? | ✓ |
| Run code | ? | ✓ |
| Use a computer | ? | ✓ |
What breaks when you switch
Opus 5 → Sonnet 5.5
No forced tool use›
Forced tool_choice any or tool returns HTTP 400; remove the forced choice.
Replace disabled thinking›
Disabled thinking returns HTTP 400; use between_tools for the lowest thinking setting.
No manual thinking budgets›
Manual thinking budgets and extra between_tools fields return HTTP 400; remove manual budgets and send between_tools without display, budget_tokens or block_binding.
Remove sampling parameters›
Non-default temperature, top_p or top_k returns HTTP 400; remove these parameters.
No assistant prefill›
A prefilled final assistant turn returns HTTP 400; remove that prefill.
Old computer tool rejected · Claude API / Google Cloud›
The computer_20251124 tool returns HTTP 400 on the Claude API and Google Cloud; use computer_toolset_20260801 there.
Sonnet 5.5 → Opus 5
Nothing known.