Claude Fable 5 vs GPT-5.6 Luna
Wins 7 of 8 areas
Coding · Agents · Reasoning · Facts · Images and charts · Math · Following instructions
Wins 1 of 8 areas
Long documents
Claude Fable 5 is the stronger all-rounder.GPT-5.6 Luna is cheaper and better at long documents.
Scores updated · 51 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.
- CodingWriting and fixing software70Claude Fable 57 of 7 tests
- AgentsCarrying out multi-step tasks on its own60Claude Fable 56 of 6 tests
- ReasoningHard problems that need careful thinking60Claude Fable 56 of 6 tests
- FactsGetting facts right instead of making them up30Claude Fable 53 of 3 tests
- MathCompetition and research-level math30Claude Fable 53 of 3 tests
- Images and chartsUnderstanding pictures, charts and video20Claude Fable 52 of 2 tests
- Following instructionsDoing exactly what it is asked10Claude Fable 51 of 1 test
- Long documentsFinding answers in very long texts01GPT-5.6 Luna1 of 1 test
Long documents and following instructions rest on a single test each.
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 Luna costs 98% less for the same work.
The biggest differences
The tests each model wins by the widest margin, up to three each. Scores are out of 100.
Where Claude Fable 5 pulls ahead
- Common-sense trick questionsSimpleBench+35.1points ahead
- Extremely hard research-level math problemsFrontierMath Tier 4+31.7points ahead
- Complex command-line tasks across many fieldsTerminal-Bench 4.0+30.8points ahead
Where GPT-5.6 Luna pulls ahead
- Reasons across sets of long documentsAA-LCR+1.4points ahead
Every test, side by side
All 51 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 17.38062.7+17.3
- LiveBench · Agentic CodingClaude Fable 5 by 13.862.248.4+13.8
- LMArena · WebDevClaude Fable 5 by 106 rating points16251519+106 rating
- SciCodeClaude Fable 5 by 7.46153.6+7.4
- Terminal-Bench 2.1Claude Fable 5 by 3.784.680.9+3.7
- LiveBench · CodingClaude Fable 5 by 3.18682.9+3.1
- SWE-bench VerifiedClaude Fable 5 by 29593+2
AgentsClaude Fable 5
- Terminal-Bench 4.0Claude Fable 5 by 30.842.411.6+30.8
- OSWorld-VerifiedClaude Fable 5 by 12.48572.6+12.4
- AA ApexAgentsClaude Fable 5 by 7.543.335.8+7.5
- GDPValClaude Fable 5 by 7.555.548+7.5
- τ-Bench V3 · BankingClaude Fable 5 by 738.131.1+7
- BrowseCompClaude Fable 5 by 4.187.483.3+4.1
ReasoningClaude Fable 5
- SimpleBenchClaude Fable 5 by 35.181.946.8+35.1
- ARC-AGI-2Claude Fable 5 by 29.689.259.6+29.6
- Humanity's Last ExamClaude Fable 5 by 1655.539.5+16
- CritPtClaude Fable 5 by 828.620.6+8
- LiveBench · ReasoningClaude Fable 5 by 4.189.785.6+4.1
- GPQA DiamondClaude Fable 5 by 1.592.691.1+1.5
FactsClaude Fable 5
- SimpleQA VerifiedClaude Fable 5 by 29.770.741+29.7
- AA-Omniscience · Non-hallucinationClaude Fable 5 by 2936.47.4+29
- AA-Omniscience · AccuracyClaude Fable 5 by 22.765.342.7+22.7
Images and chartsClaude Fable 5
- LMArena · VisionClaude Fable 5 by 70 rating points13261256+70 rating
- MMMU-ProClaude Fable 5 by 5.684.278.6+5.6
MathClaude Fable 5
- FrontierMath Tier 4Claude Fable 5 by 31.790.258.5+31.7
- LiveBench · MathematicsClaude Fable 5 by 8.89687.2+8.8
- FrontierMath Tiers 1-3 (v2)Claude Fable 5 by 4.98782.1+4.9
Following instructionsClaude Fable 5
- LiveBench · Instruction FollowingClaude Fable 5 by 15.775.860.1+15.7
Other results22 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.
- AA-OmniscienceClaude Fable 5 by 53.643.3-10.3+53.6
- Terminal-Bench 4.0Claude Fable 5 by 27.244.517.3+27.2
- Agents' Last ExamGPT-5.6 Luna by 24.625.750.3+24.6
- OSWorld 2.0Claude Fable 5 by 20.566.145.6+20.5
- GDPval-AA v2Claude Fable 5 by 193 rating points17231530+193 rating
- livebench_languageClaude Fable 5 by 18.190.772.6+18.1
- SWEBench Pro PublicClaude Fable 5 by 17.38062.7+17.3
- HLE (with tools)Claude Fable 5 by 14.963.848.9+14.9
- GDPval-AA v2 EloClaude Fable 5 by 139 rating points17231584+139 rating
- AA IntelligenceClaude Fable 5 by 12.349.637.3+12.3
- HealthBench ProfessionalClaude Fable 5 by 10.36655.7+10.3
- Toolathlon VerifiedClaude Fable 5 by 1077.967.9+10
- AA Agentic IndexClaude Fable 5 by 8.35142.7+8.3
- ToolathlonClaude Fable 5 by 8.361.753.4+8.3
- ARC-AGI-1Claude Fable 5 by 7.898.590.7+7.8
- GDP (Surge AI)Claude Fable 5 by 7.129.822.7+7.1
- HealthBenchClaude Fable 5 by 5.461.255.8+5.4
- Artificial Analysis Coding IndexClaude Fable 5 by 5.176.571.4+5.1
- DeepSWE 1.1Claude Fable 5 by 37067+3
- DeepSWEClaude Fable 5 by 2.87067.2+2.8
- livebench_data_analysisClaude Fable 5 by 2.580.578+2.5
- τ³-Bench BankingClaude Fable 5 by 2.526.824.3+2.5
Questions people ask
Which is better, Claude Fable 5 or GPT-5.6 Luna?
Claude Fable 5 wins seven of the eight areas we test: coding, agents, reasoning, facts, images and charts, math and following instructions. GPT-5.6 Luna wins long documents, and costs 98% less.
Which is better for coding?
Claude Fable 5. It wins 7 of the 7 coding tests both models report; GPT-5.6 Luna wins none.
Which is cheaper?
Claude Fable 5 costs $10.00 per million input tokens and $50.00 per million output tokens; GPT-5.6 Luna costs $0.20 and $1.20. That makes GPT-5.6 Luna about 98% cheaper for the same work.
How do you compare the two?
We use the 51 benchmark tests both models have published scores on. The verdict counts the 29 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 22 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.