Claude Opus 5.5 vs Claude Opus 5
Wins 8 of 8 areas
Coding · Agents · Reasoning · Facts · Images and charts · Math · Long documents · Following instructions
Wins 0 of 8 areas
—
Claude Opus 5.5 is the stronger all-rounder.
Both rank among the ten best models we track in reasoning, coding, agents 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 software60Claude Opus 5.56 of 6 tests
- ReasoningHard problems that need careful thinking50Claude Opus 5.55 of 5 tests
- FactsGetting facts right instead of making them up30Claude Opus 5.53 of 3 tests
- MathCompetition and research-level math30Claude Opus 5.53 of 3 tests
- AgentsCarrying out multi-step tasks on its own20Claude Opus 5.52 of 3 tests · 1 tie
- Images and chartsUnderstanding pictures, charts and video10Claude Opus 5.51 of 2 tests · 1 tie
- Long documentsFinding answers in very long texts10Claude Opus 5.51 of 1 test
- Following instructionsDoing exactly what it is asked10Claude Opus 5.51 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.
Claude Opus 5.5 costs 20% 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 Opus 5.5 pulls ahead
- Extremely hard research-level math problemsFrontierMath Tier 4+21.8points ahead
- Short factual questions, answered correctlySimpleQA Verified+12.3points ahead
- Long, multi-file coding tasks in real codebasesSWE-bench Pro+10.7points ahead
Where Claude Opus 5 pulls ahead
No clear win on a test scored out of 100.
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 Opus 5.5
- LMArena · WebDevClaude Opus 5.5 by 123 rating points18141691+123 rating
- SWE-bench ProClaude Opus 5.5 by 10.789.979.2+10.7
- SciCodeClaude Opus 5.5 by 10.566.956.4+10.5
- LiveBench · CodingClaude Opus 5.5 by 7.889.381.5+7.8
- LiveBench · Agentic CodingClaude Opus 5.5 by 6.571.765.2+6.5
- SWE-bench MultilingualClaude Opus 5.5 by 4.493.989.5+4.4
AgentsClaude Opus 5.5
- Terminal-Bench 4.0Claude Opus 5.5 by 10.659.649+10.6
- GDPValClaude Opus 5.5 by 7.268.361.1+7.2
- BrowseComptie90.890.8tie
ReasoningClaude Opus 5.5
- SimpleBenchClaude Opus 5.5 by 7.888.480.6+7.8
- Humanity's Last ExamClaude Opus 5.5 by 6.561.454.9+6.5
- CritPtClaude Opus 5.5 by 2.631.729.1+2.6
- ARC-AGI-2Claude Opus 5.5 by 1.391.790.4+1.3
- LiveBench · ReasoningClaude Opus 5.5 by 192.291.2+1
FactsClaude Opus 5.5
- SimpleQA VerifiedClaude Opus 5.5 by 12.372.259.9+12.3
- AA-Omniscience · AccuracyClaude Opus 5.5 by 5.366.260.9+5.3
- AA-Omniscience · Non-hallucinationClaude Opus 5.5 by 2.241.439.2+2.2
Images and chartsClaude Opus 5.5
- MMMU-ProClaude Opus 5.5 by 387.784.7+3
- LMArena · Visiontie13241317tie
MathClaude Opus 5.5
- FrontierMath Tier 4Claude Opus 5.5 by 21.89573.2+21.8
- FrontierMath Tiers 1-3 (v2)Claude Opus 5.5 by 5.691.285.6+5.6
- LiveBench · MathematicsClaude Opus 5.5 by 1.497.195.7+1.4
Following instructionsClaude Opus 5.5
- LiveBench · Instruction FollowingClaude Opus 5.5 by 1.965.763.8+1.9
Other results31 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.
- Terminal-Bench-Science 0.1Claude Opus 5.5 by 305929+30
- Terminal-Bench 4.0Claude Opus 5.5 by 14.666.451.8+14.6
- GDPval-AA 2.1Claude Opus 5.5 by 138 rating points18461708+138 rating
- GDPval-AA (Elo)Claude Opus 5.5 by 112 rating points18201708+112 rating
- OSWorld 2.0Claude Opus 5.5 by 11.281.870.6+11.2
- AA-OmniscienceClaude Opus 5.5 by 9.346.437.1+9.3
- OSWorld 2.0 (strict)Claude Opus 5.5 by 9.148.739.6+9.1
- LVBenchClaude Opus 5.5 by 8.383.775.4+8.3
- OSWorld 2.1Claude Opus 5.5 by 7.881.874+7.8
- OSWorld 2.1 partialClaude Opus 5.5 by 7.881.874+7.8
- AA IntelligenceClaude Opus 5.5 by 6.857.650.8+6.8
- HealthBenchClaude Opus 5 by 6.560.667.1+6.5
- OSWorld 2.0 (partial)Claude Opus 5.5 by 6.481.875.4+6.4
- HealthBench ProfessionalClaude Opus 5.5 by 5.865.659.8+5.8
- Program BenchClaude Opus 5.5 by 5.891.285.4+5.8
- livebench_data_analysisClaude Opus 5.5 by 5.780.374.5+5.7
- DeepSWE 1.1Claude Opus 5.5 by 5.474.268.8+5.4
- HLE (with tools)Claude Opus 5.5 by 4.167.763.6+4.1
- HealthBench Professional (raw score)Claude Opus 5.5 by 3.777.173.4+3.7
- Legal Agent BenchmarkClaude Opus 5 by 3.48.311.7+3.4
- HealthBench length-adjustedClaude Opus 5.5 by 2.860.657.8+2.8
- Toolathlon VerifiedClaude Opus 5 by 2.877.880.6+2.8
- livebench_languageClaude Opus 5 by 2.486.388.7+2.4
- HiL-Bench (Tools-allowed)Claude Opus 5.5 by 25957+2
- SWE-bench MultimodalClaude Opus 5.5 by 261.459.4+2
- GMMLUClaude Opus 5.5 by 1.894.392.5+1.8
- HealthBench (raw score)Claude Opus 5.5 by 168.167.1+1
- MILUClaude Opus 5.5 by 193.192.1+1
- OfficeQAtie78.978.1tie
- OfficeQA Protie67.766.9tie
- ARC-AGI-1tie97.597.5tie
Questions people ask
Which is better, Claude Opus 5.5 or Claude Opus 5?
Claude Opus 5.5 wins all eight areas we test: coding, agents, reasoning, facts, images and charts, math, long documents and following instructions. Claude Opus 5 wins none.
Which is better for coding?
Claude Opus 5.5. It wins 6 of the 6 coding tests both models report; Claude Opus 5 wins none.
Which is cheaper?
Claude Opus 5.5 costs $4.00 per million input tokens and $20.00 per million output tokens; Claude Opus 5 costs $5.00 and $25.00. That makes Claude Opus 5.5 about 20% cheaper for the same work.
How do you compare the two?
We use the 55 benchmark tests both models have published scores on. The verdict counts the 24 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 31 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.