Compare · ModelsLive · 3 picked · head to head
Claude Opus 4.8 vs Claude Opus 5 vs GPT-5.6 Sol
Side by side · benchmarks, pricing, and signals you can act on.
Winner summary
Claude Opus 5 wins on 16/24 benchmarks
Claude Opus 5 wins 16 of 24 shared benchmarks. Leads in agentic · reasoning · coding.
Category leads
agentic·Claude Opus 5reasoning·Claude Opus 5knowledge·GPT-5.6 Solcoding·Claude Opus 5general·Claude Opus 5math·GPT-5.6 Sol
Hype vs Reality
Attention vs performance
Claude Opus 4.8
#75 by perf·#9 by attention
Claude Opus 5
#21 by perf·#9 by attention
GPT-5.6 Sol
#30 by perf·#4 by attention
Best value
GPT-5.6 Sol
2.4x better value than Claude Opus 5
Claude Opus 4.8
3.7 pts/$
$15.00/M
Claude Opus 5
4.6 pts/$
$15.00/M
GPT-5.6 Sol
11.1 pts/$
$6.00/M
Vendor risk
Who is behind the model
Anthropic
$965.0B·Tier 1
Anthropic
$965.0B·Tier 1
OpenAI
$840.0B·Tier 1
Head to head
24 benchmarks · 3 models
Claude Opus 4.8Claude Opus 5GPT-5.6 Sol
APEX-Agents
Claude Opus 5 leads by +14.4
APEX-Agents · evaluates AI agents on complex, multi-step tasks requiring planning, tool use, and autonomous decision-making in realistic environments.
Claude Opus 4.8
48.9
Claude Opus 5
65.8
GPT-5.6 Sol
51.4
ARC-AGI
ARC-AGI · the original Abstraction and Reasoning Corpus, testing whether AI can solve novel visual pattern recognition tasks without memorization.
Claude Opus 4.8
92.5
Claude Opus 5
97.5
GPT-5.6 Sol
97.5
ARC-AGI-2
GPT-5.6 Sol leads by +2.1
ARC-AGI-2 · the second iteration of the Abstraction and Reasoning Corpus, testing novel pattern recognition and abstract reasoning without prior training data.
Claude Opus 4.8
72.1
Claude Opus 5
90.4
GPT-5.6 Sol
92.5
Chess Puzzles
GPT-5.6 Sol leads by +23.1
Chess Puzzles · tests strategic and tactical reasoning by having models solve chess puzzle positions, evaluating lookahead and pattern recognition abilities.
Claude Opus 4.8
30.6
Claude Opus 5
39.0
GPT-5.6 Sol
62.1
Deepswe
Claude Opus 5 leads by +1.0
Claude Opus 4.8
59.0
Claude Opus 5
73.7
GPT-5.6 Sol
72.7
Dtbench
Claude Opus 5 leads by +3.1
Claude Opus 4.8
91.5
Claude Opus 5
96.5
GPT-5.6 Sol
93.3
Ebr Bench
Claude Opus 5 leads by +1.0
Claude Opus 4.8
28.6
Claude Opus 5
45.7
GPT-5.6 Sol
44.8
Frontiercode
Claude Opus 5 leads by +5.9
Claude Opus 4.8
46.5
Claude Opus 5
53.4
GPT-5.6 Sol
47.5
FrontierMath-Tier-4-v2-Private
GPT-5.6 Sol leads by +9.8
Claude Opus 4.8
56.1
Claude Opus 5
73.2
GPT-5.6 Sol
82.9
FrontierMath-Tiers-1-3-v2-Private
GPT-5.6 Sol leads by +3.5
Claude Opus 4.8
80.0
Claude Opus 5
85.6
GPT-5.6 Sol
89.1
Furniture Assembly
Claude Opus 5 leads by +5.9
Claude Opus 4.8
17.9
Claude Opus 5
44.0
GPT-5.6 Sol
38.1
GPQA diamond
Claude Opus 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 4.8
88.0
Claude Opus 5
91.8
GPT-5.6 Sol
91.3
Lmca
Claude Opus 5 leads by +6.2
Claude Opus 4.8
67.7
Claude Opus 5
75.8
GPT-5.6 Sol
69.7
Mystery Game Puzzles
Claude Opus 5 leads by +1.1
Claude Opus 4.8
29.5
Claude Opus 5
54.8
GPT-5.6 Sol
53.7
Osworld 2 0
Claude Opus 5 leads by +4.1
Claude Opus 4.8
20.6
Claude Opus 5
31.4
GPT-5.6 Sol
27.3
OTIS Mock AIME 2024-2025
GPT-5.6 Sol leads by +1.1
OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills.
Claude Opus 4.8
98.3
Claude Opus 5
98.9
GPT-5.6 Sol
100.0
PostTrainBench
GPT-5.6 Sol leads by +1.2
Claude Opus 4.8
33.8
Claude Opus 5
35.0
GPT-5.6 Sol
36.2
Proofbench
Claude Opus 5 leads by +16.0
Claude Opus 4.8
69.0
Claude Opus 5
99.0
GPT-5.6 Sol
83.0
SimpleBench
Claude Opus 5 leads by +10.7
SimpleBench · tests fundamental reasoning capabilities with straightforward problems designed to expose gaps in basic logical and spatial thinking.
Claude Opus 4.8
57.8
Claude Opus 5
76.7
GPT-5.6 Sol
66.0
SimpleQA Verified
GPT-5.6 Sol leads by +9.8
SimpleQA Verified · short factual questions with verified answers, measuring factual accuracy and the tendency to hallucinate or provide incorrect information.
Claude Opus 4.8
53.0
Claude Opus 5
59.9
GPT-5.6 Sol
69.7
WeirdML
Claude Opus 5 leads by +2.3
WeirdML · tests models on unusual and adversarial machine learning tasks that require creative problem-solving beyond standard patterns.
Claude Opus 4.8
82.9
Claude Opus 5
91.8
GPT-5.6 Sol
89.4
Balrog
Claude Opus 5 leads by +3.4
Balrog · benchmarks AI agents on text-based adventure games, testing language understanding, strategic planning, and long-horizon reasoning.
Claude Opus 5
63.4
GPT-5.6 Sol
60.0
Frontierswe
Claude Opus 5 leads by +19.8
Claude Opus 5
52.0
GPT-5.6 Sol
32.2
Surface Evolver Bench
GPT-5.6 Sol leads by +5.6
Claude Opus 4.8
87.5
GPT-5.6 Sol
93.1
Full benchmark table
| Benchmark | Claude Opus 4.8 | Claude Opus 5 | GPT-5.6 Sol |
|---|---|---|---|
APEX-Agents APEX-Agents · evaluates AI agents on complex, multi-step tasks requiring planning, tool use, and autonomous decision-making in realistic environments. | 48.9 | 65.8 | 51.4 |
ARC-AGI ARC-AGI · the original Abstraction and Reasoning Corpus, testing whether AI can solve novel visual pattern recognition tasks without memorization. | 92.5 | 97.5 | 97.5 |
ARC-AGI-2 ARC-AGI-2 · the second iteration of the Abstraction and Reasoning Corpus, testing novel pattern recognition and abstract reasoning without prior training data. | 72.1 | 90.4 | 92.5 |
Chess Puzzles Chess Puzzles · tests strategic and tactical reasoning by having models solve chess puzzle positions, evaluating lookahead and pattern recognition abilities. | 30.6 | 39.0 | 62.1 |
Deepswe | 59.0 | 73.7 | 72.7 |
Dtbench | 91.5 | 96.5 | 93.3 |
Ebr Bench | 28.6 | 45.7 | 44.8 |
Frontiercode | 46.5 | 53.4 | 47.5 |
FrontierMath-Tier-4-v2-Private | 56.1 | 73.2 | 82.9 |
FrontierMath-Tiers-1-3-v2-Private | 80.0 | 85.6 | 89.1 |
Furniture Assembly | 17.9 | 44.0 | 38.1 |
GPQA diamond Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs. | 88.0 | 91.8 | 91.3 |
Lmca | 67.7 | 75.8 | 69.7 |
Mystery Game Puzzles | 29.5 | 54.8 | 53.7 |
Osworld 2 0 | 20.6 | 31.4 | 27.3 |
OTIS Mock AIME 2024-2025 OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills. | 98.3 | 98.9 | 100.0 |
PostTrainBench | 33.8 | 35.0 | 36.2 |
Proofbench | 69.0 | 99.0 | 83.0 |
SimpleBench SimpleBench · tests fundamental reasoning capabilities with straightforward problems designed to expose gaps in basic logical and spatial thinking. | 57.8 | 76.7 | 66.0 |
SimpleQA Verified SimpleQA Verified · short factual questions with verified answers, measuring factual accuracy and the tendency to hallucinate or provide incorrect information. | 53.0 | 59.9 | 69.7 |
WeirdML WeirdML · tests models on unusual and adversarial machine learning tasks that require creative problem-solving beyond standard patterns. | 82.9 | 91.8 | 89.4 |
Balrog Balrog · benchmarks AI agents on text-based adventure games, testing language understanding, strategic planning, and long-horizon reasoning. | — | 63.4 | 60.0 |
Frontierswe | — | 52.0 | 32.2 |
Surface Evolver Bench | 87.5 | — | 93.1 |
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 | |
| $5.00 | $25.00 | 1.0M tokens (~500 books) | $100.00 | |
| $2.00 | $10.00 | 1.1M tokens (~525 books) | $40.00 |