Claude Fable 5.2 vs Claude Opus 5
Wins 6 of 6 areas
Coding · Agents · Reasoning · Facts · Math · Long documents
Wins 0 of 6 areas
—
Claude Fable 5.2 is the stronger all-rounder.Claude Opus 5 is cheaper.
Both rank among the ten best models we track in math.
Scores updated · 30 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.
- ReasoningHard problems that need careful thinking40Claude Fable 5.24 of 4 tests
- CodingWriting and fixing software30Claude Fable 5.23 of 3 tests
- MathCompetition and research-level math20Claude Fable 5.22 of 2 tests
- AgentsCarrying out multi-step tasks on its own10Claude Fable 5.21 of 1 test
- FactsGetting facts right instead of making them up10Claude Fable 5.21 of 1 test
- Long documentsFinding answers in very long texts10Claude Fable 5.21 of 1 test
Agents, facts and long documents 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.
Claude Opus 5 costs 38% 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 5.2 pulls ahead
- Solves new puzzle games it has never seenARC-AGI-3+49.1points ahead
- Short factual questions, answered correctlySimpleQA Verified+34.1points ahead
- Extremely hard research-level math problemsFrontierMath Tier 4+26.8points ahead
Where Claude Opus 5 pulls ahead
No clear win on a test scored out of 100.
Every test, side by side
All 30 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingClaude Fable 5.2
- SWE-bench ProClaude Fable 5.2 by 13.89379.2+13.8
- SciCodeClaude Fable 5.2 by 11.367.756.4+11.3
- SWE-bench MultilingualClaude Fable 5.2 by 7.196.689.5+7.1
ReasoningClaude Fable 5.2
- ARC-AGI-3Claude Fable 5.2 by 49.179.330.2+49.1
- Humanity's Last ExamClaude Fable 5.2 by 19.474.354.9+19.4
- CritPtClaude Fable 5.2 by 6.335.429.1+6.3
- ARC-AGI-2Claude Fable 5.2 by 2.192.590.4+2.1
MathClaude Fable 5.2
- FrontierMath Tier 4Claude Fable 5.2 by 26.810073.2+26.8
- FrontierMath Tiers 1-3 (v2)Claude Fable 5.2 by 9.595.185.6+9.5
Other results18 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 Opus 5 by 61.324.285.5+61.3
- Terminal-Bench-Science 0.1Claude Fable 5.2 by 44.373.329+44.3
- Terminal-Bench 4.0Claude Fable 5.2 by 15.867.651.8+15.8
- OSWorld 2.1 partialClaude Fable 5.2 by 15.389.374+15.3
- SWE-bench MultimodalClaude Fable 5.2 by 14.774.159.4+14.7
- GDPval-AA 2.1Claude Fable 5.2 by 123 rating points18311708+123 rating
- HLE (with tools)Claude Fable 5.2 by 10.674.263.6+10.6
- HealthBench ProfessionalClaude Fable 5.2 by 7.967.759.8+7.9
- HealthBench Professional length-adjustedClaude Fable 5.2 by 7.967.759.8+7.9
- HealthBenchClaude Opus 5 by 7.559.667.1+7.5
- OfficeQA ProClaude Fable 5.2 by 672.966.9+6
- DeepSWE 1.1Claude Fable 5.2 by 5.574.368.8+5.5
- OfficeQAClaude Fable 5.2 by 5.383.478.1+5.3
- GMMLUClaude Fable 5.2 by 3.59692.5+3.5
- MILUClaude Fable 5.2 by 294.192.1+2
- HealthBench length-adjustedClaude Fable 5.2 by 1.859.657.8+1.8
- Program Benchtie85.685.4tie
- ARC-AGI-1tie97.597.5tie
Questions people ask
Which is better, Claude Fable 5.2 or Claude Opus 5?
Claude Fable 5.2 wins all six areas where both have results: coding, agents, reasoning, facts, math and long documents. Claude Opus 5 wins none, but costs 38% less.
Which is better for coding?
Claude Fable 5.2. It wins 3 of the 3 coding tests both models report; Claude Opus 5 wins none.
Which is cheaper?
Claude Fable 5.2 costs $8.00 per million input tokens and $40.00 per million output tokens; Claude Opus 5 costs $5.00 and $25.00. That makes Claude Opus 5 about 38% cheaper for the same work.
How do you compare the two?
We use the 30 benchmark tests both models have published scores on. The verdict counts the 12 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 18 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.