Claude Opus 4.5 vs Claude Sonnet 4.6

Verdict

Claude Sonnet 4.6 leads 8–6 across 14 shared benchmarks.

Claude Opus 4.5 6 · Claude Sonnet 4.6 8 · higher isn't always better — check the caveats below.

Per-benchmark head-to-head

Reported scores — protocols may differ. How we compare

Sort

Claude Opus 4.5Claude Sonnet 4.6

20.7
AAPEX-Agents
23.7
80
BARC-AGI
86.5
37.64
AARC-AGI-2
60.42
7.41
AChess Puzzles
8.46
81.4
BGPQA diamond
83.16
14.08
AMystery Game Puzzles
7.47
66.3
BOSWorld
72.1
86.1
AOTIS Mock AIME 2024-2025
85.78
17.29
APostTrainBench
16.42
36
AProofBench
45
41.8
ASimpleQA Verified
29
76.65
BSWE-Bench verified
75.21
63.1
BTerminal Bench
53.4
63.7
AWeirdML
66.07

Capabilities

Reasoning

Reasoning: Claude Sonnet 4.6 leads 2–1 across 3 shared reasoning benchmarks (largest gap: SimpleQA Verified, 41.8 vs 29 — single evaluator).

4 reproduced2 unverified

Agentic

Agentic: Claude Sonnet 4.6 leads 2–0 across 2 shared agentic benchmarks (largest gap: OSWorld, 72.1 vs 66.3 — single evaluator).

4 unverified

Unknown

Unknown: Evenly split 1–1 across 2 shared unknown benchmarks (largest gap: ProofBench, 45 vs 36 — single evaluator).

2 reproduced2 unverified

A higher score does not always mean a better model. Each number is task performance under a disclosed harness. Of the 28 measurements across 14 shared benchmarks, 28 single-evaluator or undisclosed-protocol reports; 43% are independently reproduced. Check any benchmark's integrity grade before reading a win as decisive.

Frequently asked questions

Claude Sonnet 4.6 leads 8–6 across 14 shared benchmarks. Claude Opus 4.5 leads 6 benchmarks and Claude Sonnet 4.6 leads 8. Higher isn't always better — see the integrity caveats on each benchmark.

Claude Opus 4.5 and Claude Sonnet 4.6 have 14 benchmarks in common in our data — those are the only rows where a direct, apples-to-apples comparison is drawn.

Claude Sonnet 4.6 leads agentic 2–0. Claude Sonnet 4.6 leads reasoning 2–1.

Follow the record

These models

Same tier