Compare · ModelsLive · 3 picked · head to head
Claude Fable 5 vs Claude Opus 5.5 vs GPT-5.4
Side by side · benchmarks, pricing, and signals you can act on.
Winner summary
Claude Opus 5.5 wins on 14/26 benchmarks
Claude Opus 5.5 wins 14 of 26 shared benchmarks. Leads in agentic · general · math.
Category leads
speed·Claude Fable 5agentic·Claude Opus 5.5reasoning·Claude Fable 5general·Claude Opus 5.5math·Claude Opus 5.5knowledge·GPT-5.4coding·Claude Opus 5.5arena·Claude Fable 5
Hype vs Reality
Attention vs performance
Claude Fable 5
#27 by perf·#12 by attention
Claude Opus 5.5
#3 by perf·#9 by attention
GPT-5.4
#104 by perf·#4 by attention
Best value
Claude Opus 5.5
1.1x better value than GPT-5.4
Claude Fable 5
2.3 pts/$
$30.00/M
Claude Opus 5.5
6.9 pts/$
$12.00/M
GPT-5.4
6.0 pts/$
$8.75/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
26 benchmarks · 3 models
Claude Fable 5Claude Opus 5.5GPT-5.4
Artificial Analysis · Quality Index
Claude Fable 5 leads by +2.2
Claude Fable 5
59.9
Claude Opus 5.5
57.6
GPT-5.4
57.2
APEX-Agents
Claude Opus 5.5 leads by +9.9
APEX-Agents · evaluates AI agents on complex, multi-step tasks requiring planning, tool use, and autonomous decision-making in realistic environments.
Claude Fable 5
63.6
Claude Opus 5.5
73.5
GPT-5.4
52.4
ARC-AGI
Claude Fable 5 leads by +1.0
ARC-AGI · the original Abstraction and Reasoning Corpus, testing whether AI can solve novel visual pattern recognition tasks without memorization.
Claude Fable 5
98.5
Claude Opus 5.5
97.5
GPT-5.4
93.7
ARC-AGI-2
Claude Opus 5.5 leads by +3.3
ARC-AGI-2 · the second iteration of the Abstraction and Reasoning Corpus, testing novel pattern recognition and abstract reasoning without prior training data.
Claude Fable 5
89.2
Claude Opus 5.5
92.5
GPT-5.4
74.0
Dtbench
Claude Opus 5.5 leads by +0.9
Claude Fable 5
97.3
Claude Opus 5.5
98.2
GPT-5.4
90.7
Ebr Bench
Claude Opus 5.5 leads by +31.9
Claude Fable 5
39.5
Claude Opus 5.5
71.4
GPT-5.4
25.4
FrontierMath-Tier-4-v2-Private
Claude Opus 5.5 leads by +4.8
Claude Fable 5
90.2
Claude Opus 5.5
95.0
GPT-5.4
49.0
FrontierMath-Tiers-1-3-v2-Private
Claude Opus 5.5 leads by +4.2
Claude Fable 5
87.0
Claude Opus 5.5
91.2
GPT-5.4
78.6
Furniture Assembly
Claude Opus 5.5 leads by +65.5
Claude Fable 5
8.3
Claude Opus 5.5
76.2
GPT-5.4
10.7
GPQA diamond
GPT-5.4 leads by +3.6
Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs.
Claude Fable 5
81.1
Claude Opus 5.5
87.5
GPT-5.4
91.1
Lmca
Claude Opus 5.5 leads by +8.3
Claude Fable 5
71.9
Claude Opus 5.5
80.3
GPT-5.4
61.1
Mirrorcode
Claude Opus 5.5 leads by +13.5
Claude Fable 5
63.9
Claude Opus 5.5
77.4
GPT-5.4
15.6
Mystery Game Puzzles
Claude Opus 5.5 leads by +20.9
Claude Fable 5
47.1
Claude Opus 5.5
68.0
GPT-5.4
30.6
OTIS Mock AIME 2024-2025
OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills.
Claude Fable 5
100.0
Claude Opus 5.5
100.0
GPT-5.4
97.8
Proofbench
Claude Opus 5.5 leads by +5.0
Claude Fable 5
95.0
Claude Opus 5.5
100.0
GPT-5.4
56.0
SimpleQA Verified
Claude Opus 5.5 leads by +1.5
SimpleQA Verified · short factual questions with verified answers, measuring factual accuracy and the tendency to hallucinate or provide incorrect information.
Claude Fable 5
70.7
Claude Opus 5.5
72.2
GPT-5.4
45.1
Artificial Analysis · Agentic Index
GPT-5.4 leads by +16.6
Artificial Analysis Agentic Index · a composite score measuring how well a model performs in agentic workflows · multi-step tool use, planning, error recovery, and autonomous task completion. Aggregates results from multiple agentic benchmarks including SWE-bench, tool-use tests, and planning evaluations. The canonical single-number metric for "how good is this model as an agent?"
Claude Fable 5
52.8
GPT-5.4
69.4
Artificial Analysis · Coding Index
Claude Fable 5 leads by +19.2
Artificial Analysis Coding Index · a composite score that aggregates performance across multiple coding benchmarks into a single index. Tracks code generation quality, debugging ability, multi-language competence, and real-world software engineering tasks. Used by Artificial Analysis to rank model coding capability in a normalized, comparable format. Useful for developers choosing between models for coding-heavy workloads.
Claude Fable 5
76.5
GPT-5.4
57.3
Chatbot Arena Elo · Coding
Claude Fable 5 leads by +258.9
Claude Fable 5
1653.9
GPT-5.4
1395.0
Chatbot Arena Elo · Overall
Claude Fable 5 leads by +42.8
Claude Fable 5
1507.6
GPT-5.4
1464.8
Chess Puzzles
GPT-5.4 leads by +3.2
Chess Puzzles · tests strategic and tactical reasoning by having models solve chess puzzle positions, evaluating lookahead and pattern recognition abilities.
Claude Fable 5
37.9
GPT-5.4
41.1
Deepswe
Claude Fable 5 leads by +18.1
Claude Fable 5
69.9
GPT-5.4
51.8
Frontiercode
Claude Opus 5.5 leads by +1.2
Claude Fable 5
53.5
Claude Opus 5.5
54.6
Frontierswe
Claude Opus 5.5 leads by +15.4
Claude Fable 5
47.0
Claude Opus 5.5
62.3
PostTrainBench
Claude Fable 5 leads by +22.8
Claude Fable 5
41.8
GPT-5.4
19.0
WeirdML
Claude Fable 5 leads by +14.2
WeirdML · tests models on unusual and adversarial machine learning tasks that require creative problem-solving beyond standard patterns.
Claude Fable 5
91.9
GPT-5.4
77.7
Full benchmark table
| Benchmark | Claude Fable 5 | Claude Opus 5.5 | GPT-5.4 |
|---|---|---|---|
Artificial Analysis · Quality Index | 59.9 | 57.6 | 57.2 |
APEX-Agents APEX-Agents · evaluates AI agents on complex, multi-step tasks requiring planning, tool use, and autonomous decision-making in realistic environments. | 63.6 | 73.5 | 52.4 |
ARC-AGI ARC-AGI · the original Abstraction and Reasoning Corpus, testing whether AI can solve novel visual pattern recognition tasks without memorization. | 98.5 | 97.5 | 93.7 |
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. | 89.2 | 92.5 | 74.0 |
Dtbench | 97.3 | 98.2 | 90.7 |
Ebr Bench | 39.5 | 71.4 | 25.4 |
FrontierMath-Tier-4-v2-Private | 90.2 | 95.0 | 49.0 |
FrontierMath-Tiers-1-3-v2-Private | 87.0 | 91.2 | 78.6 |
Furniture Assembly | 8.3 | 76.2 | 10.7 |
GPQA diamond Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs. | 81.1 | 87.5 | 91.1 |
Lmca | 71.9 | 80.3 | 61.1 |
Mirrorcode | 63.9 | 77.4 | 15.6 |
Mystery Game Puzzles | 47.1 | 68.0 | 30.6 |
OTIS Mock AIME 2024-2025 OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills. | 100.0 | 100.0 | 97.8 |
Proofbench | 95.0 | 100.0 | 56.0 |
SimpleQA Verified SimpleQA Verified · short factual questions with verified answers, measuring factual accuracy and the tendency to hallucinate or provide incorrect information. | 70.7 | 72.2 | 45.1 |
Artificial Analysis · Agentic Index Artificial Analysis Agentic Index · a composite score measuring how well a model performs in agentic workflows · multi-step tool use, planning, error recovery, and autonomous task completion. Aggregates results from multiple agentic benchmarks including SWE-bench, tool-use tests, and planning evaluations. The canonical single-number metric for "how good is this model as an agent?" | 52.8 | — | 69.4 |
Artificial Analysis · Coding Index Artificial Analysis Coding Index · a composite score that aggregates performance across multiple coding benchmarks into a single index. Tracks code generation quality, debugging ability, multi-language competence, and real-world software engineering tasks. Used by Artificial Analysis to rank model coding capability in a normalized, comparable format. Useful for developers choosing between models for coding-heavy workloads. | 76.5 | — | 57.3 |
Chatbot Arena Elo · Coding | 1653.9 | — | 1395.0 |
Chatbot Arena Elo · Overall | 1507.6 | — | 1464.8 |
Chess Puzzles Chess Puzzles · tests strategic and tactical reasoning by having models solve chess puzzle positions, evaluating lookahead and pattern recognition abilities. | 37.9 | — | 41.1 |
Deepswe | 69.9 | — | 51.8 |
Frontiercode | 53.5 | 54.6 | — |
Frontierswe | 47.0 | 62.3 | — |
PostTrainBench | 41.8 | — | 19.0 |
WeirdML WeirdML · tests models on unusual and adversarial machine learning tasks that require creative problem-solving beyond standard patterns. | 91.9 | — | 77.7 |
Pricing · per 1M tokens · projected $/mo at 10M tokens
| Model | Input | Output | Context | Projected $/mo |
|---|---|---|---|---|
| $10.00 | $50.00 | 1.0M tokens (~500 books) | $200.00 | |
| $4.00 | $20.00 | 1.0M tokens (~500 books) | $80.00 | |
| $2.50 | $15.00 | 1.1M tokens (~525 books) | $56.25 |