Claude Fable 5 vs Claude Fable 5.1
Wins 1 of 8 areas
Following instructions
Wins 5 of 8 areas
Coding · Agents · Reasoning · Math · Long documents
Claude Fable 5.1 is the stronger all-rounder.Claude Fable 5 is better at following instructions.
Both rank among the ten best models we track in reasoning, coding, agents, facts and math.
Scores updated · 55 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 software06Claude Fable 5.16 of 7 tests · 1 tie
- ReasoningHard problems that need careful thinking05Claude Fable 5.15 of 6 tests · 1 tie
- AgentsCarrying out multi-step tasks on its own03Claude Fable 5.13 of 3 tests
- MathCompetition and research-level math12Claude Fable 5.12 of 3 tests
- Long documentsFinding answers in very long texts01Claude Fable 5.11 of 1 test
- Following instructionsDoing exactly what it is asked10Claude Fable 51 of 1 test
- FactsGetting facts right instead of making them up11Even1 each · 1 tie
- Images and chartsUnderstanding pictures, charts and video00Even0 each · 1 tie
Images and charts, 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.
The biggest differences
The three tests each model wins by the widest margin. Scores are out of 100.
Where Claude Fable 5.1 pulls ahead
- Complex command-line tasks across many fieldsTerminal-Bench 4.0+9.6points ahead
- Customer-service tasks in a simulated bankτ-Bench V3 · Banking+9.1points ahead
- Real work tasks from 44 professionsGDPVal+7.2points ahead
Where Claude Fable 5 pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+9points ahead
- Follows detailed instructions exactlyLiveBench · Instruction Following+2.8points ahead
- Extremely hard research-level math problemsFrontierMath Tier 4+2.4points ahead
Every test, side by side
All 55 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingClaude Fable 5.1
- LMArena · WebDevClaude Fable 5.1 by 116 rating points16251741+116 rating
- Terminal-Bench 2.1Claude Fable 5.1 by 6.884.691.4+6.8
- LiveBench · Agentic CodingClaude Fable 5.1 by 3.962.266.1+3.9
- SWE-bench MultilingualClaude Fable 5.1 by 2.586.689.1+2.5
- SciCodeClaude Fable 5.1 by 2.16163.1+2.1
- SWE-bench ProClaude Fable 5.1 by 1.28081.2+1.2
- LiveBench · Codingtie8686.4tie
AgentsClaude Fable 5.1
- Terminal-Bench 4.0Claude Fable 5.1 by 9.642.452+9.6
- τ-Bench V3 · BankingClaude Fable 5.1 by 9.138.147.2+9.1
- GDPValClaude Fable 5.1 by 7.255.562.7+7.2
ReasoningClaude Fable 5.1
- SimpleBenchClaude Fable 5.1 by 4.781.986.6+4.7
- Humanity's Last ExamClaude Fable 5.1 by 3.655.559.1+3.6
- LiveBench · ReasoningClaude Fable 5.1 by 289.791.7+2
- CritPtClaude Fable 5.1 by 1.128.629.7+1.1
- GPQA DiamondClaude Fable 5.1 by 1.192.693.7+1.1
- ARC-AGI-2tie89.290tie
FactsEven
- AA-Omniscience · Non-hallucinationClaude Fable 5 by 936.427.4+9
- AA-Omniscience · AccuracyClaude Fable 5.1 by 1.865.367.2+1.8
- SimpleQA Verifiedtie70.770.8tie
MathClaude Fable 5.1
- FrontierMath Tiers 1-3 (v2)Claude Fable 5.1 by 3.28790.2+3.2
- FrontierMath Tier 4Claude Fable 5 by 2.490.287.8+2.4
- LiveBench · MathematicsClaude Fable 5.1 by 19697+1
Following instructionsClaude Fable 5
- LiveBench · Instruction FollowingClaude Fable 5 by 2.875.873+2.8
Other results30 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.
- GDP (Surge AI)Claude Fable 5.1 by 55.329.885.1+55.3
- Terminal-Bench-Science 0.1Claude Fable 5.1 by 27.924.752.6+27.9
- GDPval-AA (Elo)Claude Fable 5 by 197 rating points19321735+197 rating
- GDPval-AA 2.1Claude Fable 5.1 by 140 rating points15951735+140 rating
- GDPval-AA v2Claude Fable 5.1 by 130 rating points17231853+130 rating
- OSWorld 2.0Claude Fable 5.1 by 11.866.177.9+11.8
- Terminal-Bench 4.0Claude Fable 5.1 by 11.344.555.8+11.3
- GDPval-AA v2 EloClaude Fable 5.1 by 112 rating points17231835+112 rating
- OfficeQA ProClaude Fable 5.1 by 11.157.969+11.1
- AA Agentic IndexClaude Fable 5.1 by 75158+7
- OSWorld 2.0 (strict)Claude Fable 5.1 by 5.636.141.7+5.6
- HealthBench (raw)Claude Fable 5.1 by 5.561.266.7+5.5
- HealthBench Professional (raw)Claude Fable 5.1 by 5.368.974.2+5.3
- HiL-Bench (Tools-allowed)Claude Fable 5.1 by 5.256.361.5+5.2
- Artificial Analysis Coding IndexClaude Fable 5.1 by 5.176.581.6+5.1
- OSWorld 2.0 (partial)Claude Fable 5.1 by 572.977.9+5
- HealthBench ProfessionalClaude Fable 5 by 3.96662.1+3.9
- AA IntelligenceClaude Fable 5.1 by 3.849.653.4+3.8
- DeepSWE 1.1Claude Fable 5 by 2.67067.4+2.6
- Program BenchClaude Fable 5.1 by 1.386.387.6+1.3
- HLE (with tools)Claude Fable 5.1 by 1.263.865+1.2
- livebench_languageClaude Fable 5 by 1.290.789.5+1.2
- PostTrainBenchClaude Fable 5 by 1.241.440.2+1.2
- ARC-AGI-1Claude Fable 5 by 198.597.5+1
- SWE-bench Multimodaltie54.154.7tie
- GMMLUtie93.694tie
- AA-Omnisciencetie43.343.5tie
- livebench_data_analysistie80.580.3tie
- MILUtie92.993tie
- Toolathlon Verifiedtie77.977.8tie
Questions people ask
Which is better, Claude Fable 5 or Claude Fable 5.1?
Claude Fable 5.1 wins five of the eight areas we test: coding, agents, reasoning, math and long documents. Claude Fable 5 wins following instructions. They are level on facts and images and charts.
Which is better for coding?
Claude Fable 5.1. It wins 6 of the 7 coding tests both models report; Claude Fable 5 wins none, 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; Claude Fable 5.1 costs $10.00 and $50.00. For a million tokens read plus a million written, they cost about the same.
How do you compare the two?
We use the 55 benchmark tests both models have published scores on. The verdict counts the 25 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 30 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.