Claude Opus 4.5 vs Claude Opus 4.6
Different efforts: non-thinking · high · Compare all at high
Humanity's Last Exam Humanity's Last Exam: Very hard expert questions across many fields, written to stump today’s models
Opus 4.513.2
Opus 4.619.1
Arena Text · overall Arena Text · overall: Human pairwise preferences in the stated text vote pool
Opus 4.5 · high1,474
Opus 4.61,503
Arena Text · Chinese Arena Text · chinese: Human pairwise preferences in the stated text vote pool
Opus 4.51,469.8
Opus 4.61,550.2
Arena Text · English Arena Text · english: Human pairwise preferences in the stated text vote pool
Opus 4.51,460.6
Opus 4.61,512
Arena Text · French Arena Text · french: Human pairwise preferences in the stated text vote pool
Opus 4.51,470.8
Opus 4.61,512.8
Arena Text · German Arena Text · german: Human pairwise preferences in the stated text vote pool
Opus 4.51,449.4
Opus 4.61,501.8
Arena Text · instruction following Arena Text · instruction following: Human pairwise preferences in the stated text vote pool
Opus 4.51,472.5
Opus 4.61,522
Arena Text · Japanese Arena Text · japanese: Human pairwise preferences in the stated text vote pool
Opus 4.51,411.8
Opus 4.61,488
Arena Text · Korean Arena Text · korean: Human pairwise preferences in the stated text vote pool
Opus 4.51,423.6
Opus 4.61,449.8
Arena Text · non-English Arena Text · non english: Human pairwise preferences in the stated text vote pool
Opus 4.51,437.5
Opus 4.61,488.1
Arena Text · Polish Arena Text · polish: Human pairwise preferences in the stated text vote pool
Opus 4.51,438.7
Opus 4.61,496.6
Arena Text · Russian Arena Text · russian: Human pairwise preferences in the stated text vote pool
Opus 4.51,447.4
Opus 4.61,496.1
Arena Text · Spanish Arena Text · spanish: Human pairwise preferences in the stated text vote pool
Opus 4.51,452.5
Opus 4.61,504.2
AA Omniscience Index AA Omniscience Index: Scores factual knowledge while penalizing incorrect answers to encourage admitting uncertainty
Opus 4.5-4.1
Opus 4.62.4
SimpleQA Verified v1 SimpleQA Verified v1: Accuracy on short factual questions with verified answers, answered without web access
Opus 4.539
Opus 4.6 · max47
CritPt (physics) CritPt (physics): Research-level physics problems that need many careful reasoning steps
Opus 4.50.3
Opus 4.62.8
AA Intelligence Index 4.3.2 AA Intelligence Index 4.3.2: Artificial Analysis combines tests of knowledge, reasoning, coding and tool use into one score
Opus 4.523.7
Opus 4.626.4
AA Omniscience · Accuracy AA Omniscience · Accuracy: Measures correct answers on the AA Omniscience factual knowledge test
Opus 4.540.9
Opus 4.645.8
AA Omniscience · Hallucination Rate AA Omniscience · Hallucination Rate: Measures how often attempted answers are incorrect on the AA Omniscience factual knowledge test · lower is better
Opus 4.576.2
Opus 4.680.1
AA-LCR (long context) 1.1 AA-LCR (long context) 1.1: Questions that need reading and connecting facts across very long documents (around 100K tokens)
Opus 4.570.7
Opus 4.667
IFBench (instructions) IFBench (instructions): Checks whether the model follows precise formatting and content instructions
Opus 4.543
Opus 4.644.6
EQ-Bench 3 EQ-Bench 3: Judge-rated responses to social and emotionally nuanced scenarios, scored through the source pairwise ranking
Opus 4.51,545.7
Opus 4.61,717.4
Chess Puzzles Chess Puzzles: Models choose the best next move from a novel chess position
Opus 4.55
Opus 4.6 · max21
FrontierMath FrontierMath: Models solve expert-written advanced mathematics problems on the named tier and edition
Opus 4.520.7
Opus 4.6 · max40.7
FrontierMath Tier 4 FrontierMath Tier 4: Models solve expert-written advanced mathematics problems on the named tier and edition
Opus 4.54.2
Opus 4.6 · max22.9
Vals LegalBench Vals LegalBench: Legal reasoning and legal-language understanding on LegalBench
Opus 4.584.6
Opus 4.685.3
Vals MedCode Vals MedCode: Medical coding accuracy on patient encounters
Opus 4.549.2
Opus 4.649.1
Vals MedScribe Vals MedScribe: Accuracy of clinical notes produced from consultation transcripts
Opus 4.585.3
Opus 4.686.1
ARC-AGI-1 · public evaluation ARC-AGI-1 · public-eval: Models infer transformations from examples and solve unseen grid puzzles; the evaluation split is explicit
Opus 4.5—
Opus 4.696.3
ARC-AGI-1 · semi-private ARC-AGI-1 · semi-private: Models infer transformations from examples and solve unseen grid puzzles; the evaluation split is explicit
Opus 4.5—
Opus 4.694
ARC-AGI-2 · public evaluation ARC-AGI-2 · public-eval: Models infer transformations from examples and solve unseen grid puzzles; the evaluation split is explicit
Opus 4.5—
Opus 4.679
ARC-AGI-2 · semi-private ARC-AGI-2 · semi-private: Models infer transformations from examples and solve unseen grid puzzles; the evaluation split is explicit
Opus 4.5—
Opus 4.669.2
FrontierMath Tier 4 (v2) FrontierMath Tier 4 (v2): Models solve expert-written advanced mathematics problems on the named tier and edition
Opus 4.5—
Opus 4.6 · max26.8
FrontierMath Tiers 1–3 (v2) FrontierMath Tiers 1–3 (v2): Models solve expert-written advanced mathematics problems on the named tier and edition
Opus 4.5—
Opus 4.6 · max66
Mystery Game Puzzles Mystery Game Puzzles: Models choose the best move in puzzle positions from an undisclosed game
Opus 4.5—
Opus 4.6 · max25
MathArena · Apex MathArena · apex · Apex: Answer accuracy on the named mathematics competition under the recorded effort
Opus 4.5—
Opus 4.634.5
MathArena · Apex Shortlist MathArena · apex · Apex Shortlist: Answer accuracy on the named mathematics competition under the recorded effort
Opus 4.5—
Opus 4.686.7
MathArena · ArXivMath · December 2025 MathArena · arxiv · 12/2025: Answer accuracy on the named mathematics competition under the recorded effort
Opus 4.5—
Opus 4.657.4
MathArena · ArXivMath · February 2026 MathArena · arxiv · 02/2026: Answer accuracy on the named mathematics competition under the recorded effort
Opus 4.5—
Opus 4.640.6
MathArena · ArXivMath · January 2026 MathArena · arxiv · 01/2026: Answer accuracy on the named mathematics competition under the recorded effort
Opus 4.5—
Opus 4.672.8
MathArena · ArXivMath · March 2026 MathArena · arxiv · 03/2026: Answer accuracy on the named mathematics competition under the recorded effort
Opus 4.5—
Opus 4.662.5
MathArena · BrokenArXiv · February 2026 MathArena · arxiv_false · 02/2026: Answer accuracy on the named mathematics competition under the recorded effort
Opus 4.5—
Opus 4.63.2
MathArena · BrokenArXiv · March 2026 MathArena · arxiv_false · 03/2026: Answer accuracy on the named mathematics competition under the recorded effort
Opus 4.5—
Opus 4.65.8
MathArena · Project Euler MathArena · euler · Project Euler: Answer accuracy on the named mathematics competition under the recorded effort
Opus 4.5—
Opus 4.687.7
MathArena · Kangaroo 2025 1-2 MathArena · kangaroo · Kangaroo 2025 1-2: Answer accuracy on the named mathematics competition under the recorded effort
Opus 4.5—
Opus 4.659.4
MathArena · Kangaroo 2025 11-12 MathArena · kangaroo · Kangaroo 2025 11-12: Answer accuracy on the named mathematics competition under the recorded effort
Opus 4.5—
Opus 4.687.5
MathArena · Kangaroo 2025 3-4 MathArena · kangaroo · Kangaroo 2025 3-4: Answer accuracy on the named mathematics competition under the recorded effort
Opus 4.5—
Opus 4.650
MathArena · Kangaroo 2025 5-6 MathArena · kangaroo · Kangaroo 2025 5-6: Answer accuracy on the named mathematics competition under the recorded effort
Opus 4.5—
Opus 4.658.3
MathArena · Kangaroo 2025 7-8 MathArena · kangaroo · Kangaroo 2025 7-8: Answer accuracy on the named mathematics competition under the recorded effort
Opus 4.5—
Opus 4.686.7
MathArena · Kangaroo 2025 9-10 MathArena · kangaroo · Kangaroo 2025 9-10: Answer accuracy on the named mathematics competition under the recorded effort
Opus 4.5—
Opus 4.691.7
MathArena · USAMO 2026 MathArena · usamo · USAMO 2026: Answer accuracy on the named mathematics competition under the recorded effort
Opus 4.5—
Opus 4.647
| Benchmark | Opus 4.5 | Opus 4.6 |
|---|---|---|
| Humanity's Last Exam | 13.2 | 19.1 |
| Arena Text · overall | 1,474 · high | 1,503 |
| Arena Text · Chinese | 1,469.8 | 1,550.2 |
| Arena Text · English | 1,460.6 | 1,512 |
| Arena Text · French | 1,470.8 | 1,512.8 |
| Arena Text · German | 1,449.4 | 1,501.8 |
| Arena Text · instruction following | 1,472.5 | 1,522 |
| Arena Text · Japanese | 1,411.8 | 1,488 |
| Arena Text · Korean | 1,423.6 | 1,449.8 |
| Arena Text · non-English | 1,437.5 | 1,488.1 |
| Arena Text · Polish | 1,438.7 | 1,496.6 |
| Arena Text · Russian | 1,447.4 | 1,496.1 |
| Arena Text · Spanish | 1,452.5 | 1,504.2 |
| AA Omniscience Index | -4.1 | 2.4 |
| SimpleQA Verified v1 | 39 | 47 · max |
| CritPt (physics) | 0.3 | 2.8 |
| AA Intelligence Index 4.3.2 | 23.7 | 26.4 |
| AA Omniscience · Accuracy | 40.9 | 45.8 |
| AA Omniscience · Hallucination Rate | 76.2 | 80.1 |
| AA-LCR (long context) 1.1 | 70.7 | 67 |
| IFBench (instructions) | 43 | 44.6 |
| EQ-Bench 3 | 1,545.7 | 1,717.4 |
| Chess Puzzles | 5 | 21 · max |
| FrontierMath | 20.7 | 40.7 · max |
| FrontierMath Tier 4 | 4.2 | 22.9 · max |
| Vals LegalBench | 84.6 | 85.3 |
| Vals MedCode | 49.2 | 49.1 |
| Vals MedScribe | 85.3 | 86.1 |
| ARC-AGI-1 · public evaluation | — | 96.3 |
| ARC-AGI-1 · semi-private | — | 94 |
| ARC-AGI-2 · public evaluation | — | 79 |
| ARC-AGI-2 · semi-private | — | 69.2 |
| FrontierMath Tier 4 (v2) | — | 26.8 · max |
| FrontierMath Tiers 1–3 (v2) | — | 66 · max |
| Mystery Game Puzzles | — | 25 · max |
| MathArena · Apex | — | 34.5 |
| MathArena · Apex Shortlist | — | 86.7 |
| MathArena · ArXivMath · December 2025 | — | 57.4 |
| MathArena · ArXivMath · February 2026 | — | 40.6 |
| MathArena · ArXivMath · January 2026 | — | 72.8 |
| MathArena · ArXivMath · March 2026 | — | 62.5 |
| MathArena · BrokenArXiv · February 2026 | — | 3.2 |
| MathArena · BrokenArXiv · March 2026 | — | 5.8 |
| MathArena · Project Euler | — | 87.7 |
| MathArena · Kangaroo 2025 1-2 | — | 59.4 |
| MathArena · Kangaroo 2025 11-12 | — | 87.5 |
| MathArena · Kangaroo 2025 3-4 | — | 50 |
| MathArena · Kangaroo 2025 5-6 | — | 58.3 |
| MathArena · Kangaroo 2025 7-8 | — | 86.7 |
| MathArena · Kangaroo 2025 9-10 | — | 91.7 |
| MathArena · USAMO 2026 | — | 47 |
Specs
| Spec | Opus 4.5 | Opus 4.6 |
|---|---|---|
| Input, per 1M | $5 | $5 |
| Output, per 1M | $25 | $25 |
| Speed, tok/s | 51 | 45 |
| Context | 200K | 1M |
| Weights | Closed | Closed |
| Released | Nov 2025 | Feb 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 4.5 → Opus 4.6
Nothing known.
Opus 4.6 → Opus 4.5