Claude Fable 5 vs GPT-5.6 Sol
Wins 4 of 8 areas
Coding · Agents · Facts · Images and charts
Wins 2 of 8 areas
Reasoning · Long documents
Claude Fable 5 wins more areas, narrowly.GPT-5.6 Sol is cheaper and better at reasoning.
Both rank among the ten best models we track in reasoning, coding, agents and math.
Scores updated · 71 tests both models report · How we compare
Where each one wins
Tests won in each of the eight areas we test. Each piece is one test, so a longer bar means more evidence; grey means the two scored within a point of each other.
- FactsGetting facts right instead of making them up30Claude Fable 53 of 3 tests
- CodingWriting and fixing software42Claude Fable 54 of 7 tests · 1 tie
- AgentsCarrying out multi-step tasks on its own42Claude Fable 54 of 7 tests · 1 tie
- Images and chartsUnderstanding pictures, charts and video10Claude Fable 51 of 2 tests · 1 tie
- ReasoningHard problems that need careful thinking24GPT-5.6 Sol4 of 6 tests
- Long documentsFinding answers in very long texts01GPT-5.6 Sol1 of 1 test
- MathCompetition and research-level math11Even1 each · 1 tie
- Following instructionsDoing exactly what it is asked11Even1 each
Long documents rests on a single test.
What it costs
Prices per million tokens, roughly 750,000 words. The bars show the cost of a million tokens read plus a million written.
GPT-5.6 Sol costs 60% less for the same work.
The biggest differences
The three tests each model wins by the widest margin. Scores are out of 100.
Where Claude Fable 5 pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+28.6points ahead
- Common-sense trick questionsSimpleBench+17.1points ahead
- Long, multi-file coding tasks in real codebasesSWE-bench Pro+15.4points ahead
Where GPT-5.6 Sol pulls ahead
- Follows unfamiliar, precisely checkable instructionsIFBench+9.2points ahead
- Customer-service tasks in a simulated bankτ-Bench V3 · Banking+6.2points ahead
- Research-level physics problemsCritPt+3.7points ahead
Every test, side by side
All 71 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingClaude Fable 5
- SWE-bench ProClaude Fable 5 by 15.48064.6+15.4
- LiveBench · Agentic CodingClaude Fable 5 by 662.256.2+6
- SciCodeClaude Fable 5 by 3.96157.1+3.9
- Terminal-Bench 2.1GPT-5.6 Sol by 3.484.688+3.4
- Terminal-Bench HardGPT-5.6 Sol by 362.965.9+3
- LiveBench · CodingClaude Fable 5 by 2.18683.9+2.1
- LMArena · WebDevtie16241618tie
AgentsClaude Fable 5
- τ-Bench V3 · BankingGPT-5.6 Sol by 6.238.144.3+6.2
- AA ApexAgentsClaude Fable 5 by 3.443.339.9+3.4
- BrowseCompGPT-5.6 Sol by 387.490.4+3
- Terminal-Bench 4.0Claude Fable 5 by 2.542.439.9+2.5
- OSWorld-VerifiedClaude Fable 5 by 28583+2
- MCP AtlasClaude Fable 5 by 1.184.783.6+1.1
- GDPValtie55.655.6tie
ReasoningGPT-5.6 Sol
- SimpleBenchClaude Fable 5 by 17.181.964.8+17.1
- Humanity's Last ExamClaude Fable 5 by 655.549.5+6
- CritPtGPT-5.6 Sol by 3.728.632.3+3.7
- ARC-AGI-2GPT-5.6 Sol by 3.389.292.5+3.3
- LiveBench · ReasoningGPT-5.6 Sol by 289.791.7+2
- GPQA DiamondGPT-5.6 Sol by 1.592.694.1+1.5
FactsClaude Fable 5
- AA-Omniscience · Non-hallucinationClaude Fable 5 by 28.636.47.8+28.6
- AA-Omniscience · AccuracyClaude Fable 5 by 665.359.4+6
- SimpleQA VerifiedClaude Fable 5 by 170.769.7+1
Images and chartsClaude Fable 5
- LMArena · VisionClaude Fable 5 by 46 rating points13251279+46 rating
- MMMU-Protie84.283.4tie
MathEven
- FrontierMath Tier 4Claude Fable 5 by 7.290.283+7.2
- FrontierMath Tiers 1-3 (v2)GPT-5.6 Sol by 2.18789.1+2.1
- LiveBench · Mathematicstie9696.2tie
Following instructionsEven
- IFBenchGPT-5.6 Sol by 9.263.572.7+9.2
- LiveBench · Instruction FollowingClaude Fable 5 by 3.975.871.8+3.9
Other results40 tests, not counted
Tests outside the eight areas. They are not counted above: several are summary scores built from other tests, or the same test under another name.
- Program BenchClaude Fable 5 by 61.386.325+61.3
- Agents' Last ExamGPT-5.6 Sol by 2725.752.7+27
- HiL-Bench (Tools-allowed)Claude Fable 5 by 2456.332.3+24
- AA-OmniscienceClaude Fable 5 by 21.343.322+21.3
- SWEBench Pro PublicClaude Fable 5 by 15.48064.6+15.4
- τ²-Bench Telecom (AA run)Claude Fable 5 by 13.498.585.1+13.4
- JobBenchClaude Fable 5 by 1257.445.4+12
- AndroidBenchClaude Fable 5 by 10.584.574+10.5
- OSWorld 2.0 (partial)Claude Fable 5 by 10.372.962.6+10.3
- SWE-MarathonGPT-5.6 Sol by 7.53542.5+7.5
- Terminal-Bench 4.0Claude Fable 5 by 7.244.537.3+7.2
- τ³-Bench BankingGPT-5.6 Sol by 6.226.833+6.2
- HealthBench ProfessionalClaude Fable 5 by 5.56660.5+5.5
- MCP Mark VerifiedGPT-5.6 Sol by 5.587.492.9+5.5
- OfficeQA ProGPT-5.6 Sol by 5.357.963.2+5.3
- PostTrainBenchClaude Fable 5 by 5.241.436.2+5.2
- MLS Bench Litelower is betterGPT-5.6 Sol by 3.749.946.2+3.7
- ToolathlonClaude Fable 5 by 3.761.758+3.7
- OSWorld 2.0Claude Fable 5 by 3.566.162.6+3.5
- DeepSWE 1.1GPT-5.6 Sol by 37073+3
- livebench_languageClaude Fable 5 by 390.787.7+3
- Toolathlon VerifiedClaude Fable 5 by 377.974.9+3
- DeepSWEGPT-5.6 Sol by 2.77072.7+2.7
- AA IntelligenceClaude Fable 5 by 2.649.647+2.6
- SpreadsheetBench 2Claude Fable 5 by 2.334.732.4+2.3
- ARC-AGI-1Claude Fable 5 by 298.596.5+2
- PaperBenchGPT-5.6 Sol by 1.788.890.5+1.7
- charxiv_rq_with_pythonClaude Fable 5 by 1.689.487.8+1.6
- Global-MMLU-LiteClaude Fable 5 by 1.593.391.8+1.5
- BrowseComp (w/ Ctx)GPT-5.6 Sol by 1.48889.4+1.4
- GDPval-AA v2 EloClaude Fable 5 by 13 rating points17231710+13 rating
- GDPval-AA v2Claude Fable 5 by 12 rating points17231711+12 rating
- Terminal-Bench-Science 0.1GPT-5.6 Sol by 121.422.4+1
- Artificial Analysis Coding Indextie76.577.4tie
- GDP (Surge AI)tie29.830.7tie
- GDPval-AA 2.1tie15951588tie
- HLE (with tools)tie63.864.5tie
- livebench_data_analysistie80.579.8tie
- AA Agentic Indextie5150.5tie
- StrongREJECTtie98.798.5tie
Questions people ask
Which is better, Claude Fable 5 or GPT-5.6 Sol?
Claude Fable 5 wins four of the eight areas we test: coding, agents, facts and images and charts. GPT-5.6 Sol wins reasoning and long documents, and costs 60% less. They are level on math and following instructions.
Which is better for coding?
Claude Fable 5. It wins 4 of the 7 coding tests both models report; GPT-5.6 Sol wins 2, and 1 is a tie.
Which is cheaper?
Claude Fable 5 costs $10.00 per million input tokens and $50.00 per million output tokens; GPT-5.6 Sol costs $4.00 and $20.00. That makes GPT-5.6 Sol about 60% cheaper for the same work.
How do you compare the two?
We use the 71 benchmark tests both models have published scores on. The verdict counts the 31 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 40 are listed but not counted, because several are summary scores or repeat a test. Each score is the one shown on the model's own page.