Compare · ModelsLive · 2 picked · head to head
Claude Opus 4.7 vs Claude Sonnet 5
Side by side · benchmarks, pricing, and signals you can act on.
Winner summary
Claude Opus 4.7 wins on 9/15 benchmarks
Claude Opus 4.7 wins 9 of 15 shared benchmarks. Leads in general · coding · math.
Category leads
agentic·Claude Sonnet 5knowledge·Claude Sonnet 5general·Claude Opus 4.7coding·Claude Opus 4.7math·Claude Opus 4.7reasoning·Claude Opus 4.7
Hype vs Reality
Attention vs performance
Claude Opus 4.7
#120 by perf·#9 by attention
Claude Sonnet 5
#64 by perf·#12 by attention
Best value
Claude Sonnet 5
2.8x better value than Claude Opus 4.7
Claude Opus 4.7
3.4 pts/$
$15.00/M
Claude Sonnet 5
9.5 pts/$
$6.00/M
Vendor risk
Who is behind the model
Anthropic
$965.0B·Tier 1
Anthropic
$965.0B·Tier 1
Head to head
15 benchmarks · 2 models
Claude Opus 4.7Claude Sonnet 5
APEX-Agents
Claude Sonnet 5 leads by +5.3
APEX-Agents · evaluates AI agents on complex, multi-step tasks requiring planning, tool use, and autonomous decision-making in realistic environments.
Claude Opus 4.7
49.2
Claude Sonnet 5
54.5
Chess Puzzles
Claude Sonnet 5 leads by +4.2
Chess Puzzles · tests strategic and tactical reasoning by having models solve chess puzzle positions, evaluating lookahead and pattern recognition abilities.
Claude Opus 4.7
37.9
Claude Sonnet 5
42.1
Dtbench
Claude Opus 4.7 leads by +3.6
Claude Opus 4.7
91.1
Claude Sonnet 5
87.5
Frontiercode
Claude Sonnet 5 leads by +4.2
Claude Opus 4.7
38.5
Claude Sonnet 5
42.7
FrontierMath-Tier-4-v2-Private
Claude Opus 4.7 leads by +2.4
Claude Opus 4.7
31.7
Claude Sonnet 5
29.3
FrontierMath-Tiers-1-3-v2-Private
Claude Opus 4.7 leads by +4.6
Claude Opus 4.7
70.2
Claude Sonnet 5
65.6
GPQA diamond
Claude Sonnet 5 leads by +0.5
Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs.
Claude Opus 4.7
86.9
Claude Sonnet 5
87.4
GSO-Bench
Claude Opus 4.7 leads by +6.9
GSO-Bench · evaluates AI models on real-world open-source software engineering tasks, testing the ability to understand and resolve actual GitHub issues.
Claude Opus 4.7
44.1
Claude Sonnet 5
37.3
Lmca
Claude Opus 4.7 leads by +2.6
Claude Opus 4.7
61.4
Claude Sonnet 5
58.9
Mystery Game Puzzles
Claude Sonnet 5 leads by +7.7
Claude Opus 4.7
20.7
Claude Sonnet 5
28.4
OTIS Mock AIME 2024-2025
Claude Opus 4.7 leads by +3.1
OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills.
Claude Opus 4.7
97.8
Claude Sonnet 5
94.7
Proofbench
Claude Sonnet 5 leads by +23.0
Claude Opus 4.7
54.0
Claude Sonnet 5
77.0
SimpleBench
Claude Opus 4.7 leads by +1.3
SimpleBench · tests fundamental reasoning capabilities with straightforward problems designed to expose gaps in basic logical and spatial thinking.
Claude Opus 4.7
54.0
Claude Sonnet 5
52.7
SimpleQA Verified
Claude Opus 4.7 leads by +18.0
SimpleQA Verified · short factual questions with verified answers, measuring factual accuracy and the tendency to hallucinate or provide incorrect information.
Claude Opus 4.7
51.7
Claude Sonnet 5
33.7
WeirdML
Claude Opus 4.7 leads by +7.7
WeirdML · tests models on unusual and adversarial machine learning tasks that require creative problem-solving beyond standard patterns.
Claude Opus 4.7
76.4
Claude Sonnet 5
68.8
Full benchmark table
| Benchmark | Claude Opus 4.7 | Claude Sonnet 5 |
|---|---|---|
APEX-Agents APEX-Agents · evaluates AI agents on complex, multi-step tasks requiring planning, tool use, and autonomous decision-making in realistic environments. | 49.2 | 54.5 |
Chess Puzzles Chess Puzzles · tests strategic and tactical reasoning by having models solve chess puzzle positions, evaluating lookahead and pattern recognition abilities. | 37.9 | 42.1 |
Dtbench | 91.1 | 87.5 |
Frontiercode | 38.5 | 42.7 |
FrontierMath-Tier-4-v2-Private | 31.7 | 29.3 |
FrontierMath-Tiers-1-3-v2-Private | 70.2 | 65.6 |
GPQA diamond Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs. | 86.9 | 87.4 |
GSO-Bench GSO-Bench · evaluates AI models on real-world open-source software engineering tasks, testing the ability to understand and resolve actual GitHub issues. | 44.1 | 37.3 |
Lmca | 61.4 | 58.9 |
Mystery Game Puzzles | 20.7 | 28.4 |
OTIS Mock AIME 2024-2025 OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills. | 97.8 | 94.7 |
Proofbench | 54.0 | 77.0 |
SimpleBench SimpleBench · tests fundamental reasoning capabilities with straightforward problems designed to expose gaps in basic logical and spatial thinking. | 54.0 | 52.7 |
SimpleQA Verified SimpleQA Verified · short factual questions with verified answers, measuring factual accuracy and the tendency to hallucinate or provide incorrect information. | 51.7 | 33.7 |
WeirdML WeirdML · tests models on unusual and adversarial machine learning tasks that require creative problem-solving beyond standard patterns. | 76.4 | 68.8 |
Pricing · per 1M tokens · projected $/mo at 10M tokens
| Model | Input | Output | Context | Projected $/mo |
|---|---|---|---|---|
| $5.00 | $25.00 | 1.0M tokens (~500 books) | $100.00 | |
| $2.00 | $10.00 | 1.0M tokens (~500 books) | $40.00 |
People also compared