Claude Opus 4.8 vs Claude Sonnet 5
Wins 7 of 8 areas
Coding · Agents · Reasoning · Facts · Images and charts · Math · Following instructions
Wins 1 of 8 areas
Long documents
Claude Opus 4.8 is the stronger all-rounder.Claude Sonnet 5 is cheaper and better at long documents.
Scores updated · 59 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 software61Claude Opus 4.86 of 8 tests · 1 tie
- MathCompetition and research-level math40Claude Opus 4.84 of 4 tests
- ReasoningHard problems that need careful thinking30Claude Opus 4.83 of 5 tests · 2 ties
- FactsGetting facts right instead of making them up20Claude Opus 4.82 of 3 tests · 1 tie
- Images and chartsUnderstanding pictures, charts and video20Claude Opus 4.82 of 2 tests
- AgentsCarrying out multi-step tasks on its own21Claude Opus 4.82 of 5 tests · 2 ties
- Following instructionsDoing exactly what it is asked10Claude Opus 4.81 of 1 test
- Long documentsFinding answers in very long texts01Claude Sonnet 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 Sonnet 5 costs 60% 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 Opus 4.8 pulls ahead
- Extremely hard research-level math problemsFrontierMath Tier 4+26.8points ahead
- Short factual questions, answered correctlySimpleQA Verified+19.3points ahead
- Proof problems from the US Math OlympiadUSAMO 2026+17.2points ahead
Where Claude Sonnet 5 pulls ahead
- Fixes real GitHub issues, working as an agentLiveBench · Agentic Coding+8.9points ahead
- Reasons across sets of long documentsAA-LCR+4.3points ahead
- Customer-service tasks in a simulated bankτ-Bench V3 · Banking+3.1points ahead
Every test, side by side
All 59 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingClaude Opus 4.8
- LiveBench · Agentic CodingClaude Sonnet 5 by 8.950.559.4+8.9
- SWE-bench MultilingualClaude Opus 4.8 by 6.184.478.3+6.1
- SWE-bench ProClaude Opus 4.8 by 669.263.2+6
- Terminal-Bench 2.1Claude Opus 4.8 by 4.184.680.5+4.1
- SWE-bench VerifiedClaude Opus 4.8 by 3.488.685.2+3.4
- LMArena · WebDevClaude Opus 4.8 by 15 rating points15561541+15 rating
- LiveBench · CodingClaude Opus 4.8 by 1.181.880.7+1.1
- SciCodetie54.454.3tie
AgentsClaude Opus 4.8
- Terminal-Bench 4.0Claude Opus 4.8 by 7.621.714.1+7.6
- τ-Bench V3 · BankingClaude Sonnet 5 by 3.134.237.3+3.1
- OSWorld-VerifiedClaude Opus 4.8 by 2.283.481.2+2.2
- GDPValtie47.848.3tie
- BrowseComptie84.384.7tie
ReasoningClaude Opus 4.8
- Humanity's Last ExamClaude Opus 4.8 by 7.448.741.3+7.4
- SimpleBenchClaude Opus 4.8 by 4.264.860.6+4.2
- CritPtClaude Opus 4.8 by 420.916.9+4
- GPQA Diamondtie9291.1tie
- LiveBench · Reasoningtie89.288.7tie
FactsClaude Opus 4.8
- SimpleQA VerifiedClaude Opus 4.8 by 19.35333.7+19.3
- AA-Omniscience · AccuracyClaude Opus 4.8 by 8.748.840+8.7
- AA-Omniscience · Non-hallucinationtie60.760.6tie
Images and chartsClaude Opus 4.8
- LMArena · VisionClaude Opus 4.8 by 20 rating points12951275+20 rating
- CharXiv (RQ)Claude Opus 4.8 by 1.689.988.3+1.6
MathClaude Opus 4.8
- FrontierMath Tier 4Claude Opus 4.8 by 26.856.129.3+26.8
- USAMO 2026Claude Opus 4.8 by 17.296.779.5+17.2
- FrontierMath Tiers 1-3 (v2)Claude Opus 4.8 by 14.48065.6+14.4
- LiveBench · MathematicsClaude Opus 4.8 by 1.494.392.9+1.4
Following instructionsClaude Opus 4.8
- LiveBench · Instruction FollowingClaude Opus 4.8 by 8.17263.9+8.1
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.
- AA-OmniscienceClaude Opus 4.8 by 12.328.816.4+12.3
- Terminal-Bench 4.0Claude Opus 4.8 by 11.223.612.4+11.2
- SWE-bench MultimodalClaude Opus 4.8 by 10.338.428.1+10.3
- Toolathlon Pass@∞Claude Opus 4.8 by 7.448.140.7+7.4
- OfficeQA ProClaude Opus 4.8 by 6.866.259.4+6.8
- Terminal-bench 3.0Claude Opus 4.8 by 6.521.114.6+6.5
- Terminal-Bench 2.0Claude Sonnet 5 by 5.874.680.4+5.8
- ChartMuseum (without tools)Claude Opus 4.8 by 5.775.870.1+5.7
- livebench_data_analysisClaude Sonnet 5 by 5.76671.7+5.7
- ToolathlonClaude Opus 4.8 by 5.659.954.3+5.6
- Program BenchClaude Sonnet 5 by 5.471.977.3+5.4
- Toolathlon VerifiedClaude Opus 4.8 by 5.279.974.7+5.2
- DeepSWE 1.1Claude Opus 4.8 by 55954+5
- livebench_languageClaude Opus 4.8 by 4.779.775+4.7
- Legal Agent BenchmarkClaude Opus 4.8 by 4.2105.8+4.2
- AA IntelligenceClaude Opus 4.8 by 3.641.838.2+3.6
- GDP (Surge AI)Claude Opus 4.8 by 3.284.881.6+3.2
- ChartMuseum (with tools)Claude Opus 4.8 by 389.786.7+3
- Artificial Analysis Coding IndexClaude Opus 4.8 by 2.874.371.5+2.8
- GDPval-AA v2Claude Sonnet 5 by 25 rating points15931618+25 rating
- OfficeQAClaude Opus 4.8 by 2.577.675.1+2.5
- HealthBench Professional (raw)Claude Sonnet 5 by 2.160.362.4+2.1
- HealthBench ProfessionalClaude Sonnet 5 by 255.857.8+2
- BrowseComp (multi-agent)Claude Opus 4.8 by 1.988.586.6+1.9
- AA Agentic IndexClaude Sonnet 5 by 1.742.644.3+1.7
- CharXiv Reasoning (With tools)Claude Opus 4.8 by 1.689.988.3+1.6
- GDPval-AA v2 Elotie15931584tie
- Legal Agent Benchmark, Full Public Settie9.68.9tie
- HLE (with tools)tie57.957.4tie
- HealthBench (raw)tie58.859.2tie
Questions people ask
Which is better, Claude Opus 4.8 or Claude Sonnet 5?
Claude Opus 4.8 wins seven of the eight areas we test: coding, agents, reasoning, facts, images and charts, math and following instructions. Claude Sonnet 5 wins long documents, and costs 60% less.
Which is better for coding?
Claude Opus 4.8. It wins 6 of the 8 coding tests both models report; Claude Sonnet 5 wins 1, and 1 is a tie.
Which is cheaper?
Claude Opus 4.8 costs $5.00 per million input tokens and $25.00 per million output tokens; Claude Sonnet 5 costs $2.00 and $10.00. That makes Claude Sonnet 5 about 60% cheaper for the same work.
How do you compare the two?
We use the 59 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 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.