Claude Fable 5.1 vs Sonnet 5.5
Wins 4 of 6 areas
Facts · Images and charts · Math · Following instructions
Wins 1 of 6 areas
Coding
Claude Fable 5.1 is the stronger all-rounder.Sonnet 5.5 is cheaper and better at coding.
Both rank among the ten best models we track in math.
Scores updated · 46 tests both models report · How we compare
Where each one wins
Tests won in each of the six areas where both have results. Each piece is one test, so a longer bar means more evidence; grey means the two scored within a point of each other.
- MathCompetition and research-level math20Claude Fable 5.12 of 3 tests · 1 tie
- FactsGetting facts right instead of making them up10Claude Fable 5.11 of 1 test
- Images and chartsUnderstanding pictures, charts and video10Claude Fable 5.11 of 1 test
- Following instructionsDoing exactly what it is asked10Claude Fable 5.11 of 1 test
- CodingWriting and fixing software13Sonnet 5.53 of 5 tests · 1 tie
- ReasoningHard problems that need careful thinking11Even1 each · 1 tie
Facts, images and charts 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.
Sonnet 5.5 costs 80% 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 5.1 pulls ahead
- Short factual questions, answered correctlySimpleQA Verified+24.3points ahead
- Follows detailed instructions exactlyLiveBench · Instruction Following+16.2points ahead
- Common-sense trick questionsSimpleBench+10.7points ahead
Where Sonnet 5.5 pulls ahead
- Very hard expert questions across many subjectsHumanity's Last Exam+5.4points ahead
- Programming problems, refreshed regularlyLiveBench · Coding+5points ahead
- Fixes real GitHub issues in many programming languagesSWE-bench Multilingual+1.2points ahead
Every test, side by side
All 46 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingSonnet 5.5
- LiveBench · Agentic CodingClaude Fable 5.1 by 9.866.156.3+9.8
- LiveBench · CodingSonnet 5.5 by 586.491.4+5
- LMArena · WebDevSonnet 5.5 by 37 rating points17491786+37 rating
- SWE-bench MultilingualSonnet 5.5 by 1.289.190.3+1.2
- SWE-bench Protie81.281.3tie
ReasoningEven
- SimpleBenchClaude Fable 5.1 by 10.786.675.9+10.7
- Humanity's Last ExamSonnet 5.5 by 5.459.164.5+5.4
- LiveBench · Reasoningtie91.791.6tie
Images and chartsClaude Fable 5.1
- LMArena · VisionClaude Fable 5.1 by 30 rating points13201290+30 rating
MathClaude Fable 5.1
- FrontierMath Tier 4Claude Fable 5.1 by 7.387.880.5+7.3
- FrontierMath Tiers 1-3 (v2)Claude Fable 5.1 by 1.490.288.8+1.4
- LiveBench · Mathematicstie9796.1tie
Following instructionsClaude Fable 5.1
- LiveBench · Instruction FollowingClaude Fable 5.1 by 16.27356.8+16.2
Other results32 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.
- livebench_data_analysisClaude Fable 5.1 by 20.880.359.5+20.8
- Terminal-Bench 4.0Sonnet 5.5 by 14.855.870.6+14.8
- livebench_languageClaude Fable 5.1 by 11.589.578+11.5
- GDPval-AA 2.1Sonnet 5.5 by 105 rating points17351840+105 rating
- Program BenchClaude Fable 5.1 by 7.987.679.7+7.9
- Terminal-Bench-Science 0.1Sonnet 5.5 by 7.352.659.9+7.3
- HealthBench ProfessionalSonnet 5.5 by 7.162.169.2+7.1
- HealthBench Professional length-adjustedSonnet 5.5 by 7.162.169.2+7.1
- Suicide and self-harm multi-turn appropriate response rate (Claude.ai)Sonnet 5.5 by 694100+6
- Election integrity multi-turn (appropriate response rate) - API, without a system promptClaude Fable 5.1 by 49086+4
- ArXivMath without toolsSonnet 5.5 by 3.982.986.8+3.9
- DeepSWE 1.1Sonnet 5.5 by 3.667.471+3.6
- OfficeQA ProClaude Fable 5.1 by 3.46965.6+3.4
- OfficeQAClaude Fable 5.1 by 3.380.276.9+3.3
- AA IntelligenceSonnet 5.5 by 2.653.456+2.6
- Child safety multi-turn appropriate response rate (API, without a system prompt)Sonnet 5.5 by 28486+2
- Child safety multi-turn appropriate response rate (Claude.ai)Claude Fable 5.1 by 210098+2
- Election integrity multi-turn (appropriate response rate) - Claude.aiClaude Fable 5.1 by 28886+2
- MILUClaude Fable 5.1 by 1.49391.6+1.4
- OSWorld 2.1 partialtie80.780.1tie
- HLE (with tools)tie6564.5tie
- SWE-bench Multimodaltie54.754.3tie
- Suicide and self-harm single-turn requests posing potential risk harmless rate (API, without a system prompt)tie99.399.1tie
- Suicide and self-harm single-turn requests posing potential risk harmless rate (Claude.ai)tie10099.8tie
- Suicide and self-harm single-turn benign requests refusal rate (Claude.ai)tie0.50.4tie
- Election integrity single-turn benign requests (refusal rate) - API, without a system prompttie00tie
- Election integrity single-turn benign requests (refusal rate) - Claude.aitie00tie
- Election integrity single-turn harmful requests (harmless rate) - API, without a system prompttie100100tie
- Election integrity single-turn harmful requests (harmless rate) - Claude.aitie100100tie
- Suicide and self-harm multi-turn appropriate response rate (API, without a system prompt)tie6060tie
- Suicide and self-harm single-turn benign requests refusal rate (API, without a system prompt)tie00tie
- Toolathlon Verifiedtie77.877.8tie
Questions people ask
Which is better, Claude Fable 5.1 or Sonnet 5.5?
Claude Fable 5.1 wins four of the six areas where both have results: facts, images and charts, math and following instructions. Sonnet 5.5 wins coding, and costs 80% less. They are level on reasoning.
Which is better for coding?
Sonnet 5.5. It wins 3 of the 5 coding tests both models report; Claude Fable 5.1 wins 1, and 1 is a tie.
Which is cheaper?
Claude Fable 5.1 costs $10.00 per million input tokens and $50.00 per million output tokens; Sonnet 5.5 costs $2.00 and $10.00. That makes Sonnet 5.5 about 80% cheaper for the same work.
How do you compare the two?
We use the 46 benchmark tests both models have published scores on. The verdict counts the 14 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 32 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.