Claude Opus 4.8 vs Claude Sonnet 4.6
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 4.6 is cheaper and better at long documents.
Scores updated · 69 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 software80Claude Opus 4.88 of 8 tests
- AgentsCarrying out multi-step tasks on its own60Claude Opus 4.86 of 7 tests · 1 tie
- ReasoningHard problems that need careful thinking50Claude Opus 4.85 of 5 tests
- FactsGetting facts right instead of making them up30Claude Opus 4.83 of 3 tests
- Images and chartsUnderstanding pictures, charts and video20Claude Opus 4.82 of 2 tests
- MathCompetition and research-level math20Claude Opus 4.82 of 2 tests
- Following instructionsDoing exactly what it is asked20Claude Opus 4.82 of 2 tests
- Long documentsFinding answers in very long texts01Claude Sonnet 4.61 of 1 test
Long documents rests on a single test.
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 4.6 costs 40% 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 4.8 pulls ahead
- Proof problems from the US Math OlympiadUSAMO 2026+41.7points ahead
- Multi-step tasks using many real software toolsMCP Atlas+20.9points ahead
- Short factual questions, answered correctlySimpleQA Verified+20.2points ahead
Where Claude Sonnet 4.6 pulls ahead
- Reasons across sets of long documentsAA-LCR+2.3points ahead
Every test, side by side
All 69 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingClaude Opus 4.8
- Terminal-Bench 2.1Claude Opus 4.8 by 13.484.671.2+13.4
- SWE-bench ProClaude Opus 4.8 by 11.169.258.1+11.1
- SWE-bench VerifiedClaude Opus 4.8 by 988.679.6+9
- LiveBench · Agentic CodingClaude Opus 4.8 by 7.950.542.6+7.9
- Terminal-Bench HardClaude Opus 4.8 by 5.358.353+5.3
- SciCodeClaude Opus 4.8 by 4.354.450.1+4.3
- LMArena · WebDevClaude Opus 4.8 by 34 rating points15561522+34 rating
- LiveBench · CodingClaude Opus 4.8 by 2.581.879.3+2.5
AgentsClaude Opus 4.8
- MCP AtlasClaude Opus 4.8 by 20.982.261.3+20.9
- Terminal-Bench 4.0Claude Opus 4.8 by 18.721.73+18.7
- AA ApexAgentsClaude Opus 4.8 by 11.439.428+11.4
- GDPValClaude Opus 4.8 by 11.147.836.7+11.1
- OSWorld-VerifiedClaude Opus 4.8 by 10.983.472.5+10.9
- BrowseCompClaude Opus 4.8 by 9.684.374.7+9.6
- τ-Bench V3 · Bankingtie34.234.4tie
ReasoningClaude Opus 4.8
- CritPtClaude Opus 4.8 by 17.820.93.1+17.8
- Humanity's Last ExamClaude Opus 4.8 by 15.148.733.6+15.1
- ARC-AGI-2Claude Opus 4.8 by 13.872.158.3+13.8
- GPQA DiamondClaude Opus 4.8 by 4.59287.5+4.5
- LiveBench · ReasoningClaude Opus 4.8 by 4.489.284.8+4.4
FactsClaude Opus 4.8
- SimpleQA VerifiedClaude Opus 4.8 by 20.25332.8+20.2
- AA-Omniscience · Non-hallucinationClaude Opus 4.8 by 9.160.751.6+9.1
- AA-Omniscience · AccuracyClaude Opus 4.8 by 7.948.840.9+7.9
Images and chartsClaude Opus 4.8
- CharXiv (RQ)Claude Opus 4.8 by 18.389.971.6+18.3
- LMArena · VisionClaude Opus 4.8 by 12 rating points12951283+12 rating
MathClaude Opus 4.8
- USAMO 2026Claude Opus 4.8 by 41.796.755+41.7
- LiveBench · MathematicsClaude Opus 4.8 by 7.394.387+7.3
Following instructionsClaude Opus 4.8
- LiveBench · Instruction FollowingClaude Opus 4.8 by 8.87263.2+8.8
- IFBenchClaude Opus 4.8 by 5.662.256.6+5.6
Other results39 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.
- DeepSWE 1.1Claude Opus 4.8 by 295930+29
- frontiermath_tier_4_v1Claude Opus 4.8 by 2331.38.3+23
- GDPval-AA (Elo)Claude Opus 4.8 by 214 rating points18901676+214 rating
- GDPval-AA v2Claude Opus 4.8 by 198 rating points15931395+198 rating
- ToolathlonClaude Opus 4.8 by 18.959.941+18.9
- τ²-Bench Telecom (AA run)Claude Opus 4.8 by 18.794.475.7+18.7
- AA-OmniscienceClaude Opus 4.8 by 16.628.812.2+16.6
- ChartMuseum (without tools)Claude Opus 4.8 by 16.575.859.3+16.5
- Terminal-Bench 2.0Claude Opus 4.8 by 15.574.659.1+15.5
- AgentWorldBench - MCPClaude Sonnet 4.6 by 15.154.970+15.1
- OfficeQA ProClaude Opus 4.8 by 12.866.253.4+12.8
- livebench_data_analysisClaude Sonnet 4.6 by 126678+12
- AA IntelligenceClaude Opus 4.8 by 11.741.830.1+11.7
- HealthBench ProfessionalClaude Opus 4.8 by 11.655.844.2+11.6
- Artificial Analysis Coding IndexClaude Opus 4.8 by 11.374.363+11.3
- HLE (with tools)Claude Opus 4.8 by 11.157.946.8+11.1
- Toolathlon Pass@∞Claude Opus 4.8 by 10.148.138+10.1
- AA Agentic IndexClaude Opus 4.8 by 9.542.633.1+9.5
- Finance AgentClaude Sonnet 4.6 by 9.453.963.3+9.4
- OfficeQAClaude Opus 4.8 by 8.977.668.7+8.9
- ChartMuseum (with tools)Claude Opus 4.8 by 8.889.780.9+8.8
- BrowseComp (single-agent)Claude Opus 4.8 by 8.184.376.2+8.1
- Toolathlon Avg turnsClaude Opus 4.8 by 824.516.5+8
- Blueprint-Bench 2Claude Opus 4.8 by 7.814.56.7+7.8
- ARC-AGI-1Claude Opus 4.8 by 6.592.586+6.5
- AgentWorldBench - SearchClaude Opus 4.8 by 6.335.128.8+6.3
- GDP (Surge AI)Claude Opus 4.8 by 6.284.878.6+6.2
- CharXiv Reasoning (With tools)Claude Opus 4.8 by 4.689.985.3+4.6
- Legal Agent BenchmarkClaude Opus 4.8 by 4.6105.4+4.6
- AgentWorldBench - WebClaude Opus 4.8 by 3.954.750.8+3.9
- livebench_languageClaude Opus 4.8 by 3.679.776.1+3.6
- AgentWorldBench - AndroidClaude Opus 4.8 by 3.561.558+3.5
- AgentWorldBench - OSClaude Opus 4.8 by 3.466.663.2+3.4
- Finance Agent v2Claude Opus 4.8 by 2.953.951+2.9
- AgentWorldBench - TerminalClaude Opus 4.8 by 2.259.257+2.2
- LiveBenchClaude Opus 4.8 by 1.777.275.5+1.7
- Legal Agent Benchmark, Full Public SetClaude Opus 4.8 by 1.69.68+1.6
- AgentWorldBench - Overalltie56.656tie
- AgentWorldBench - SWEtie64.164.5tie
Questions people ask
Which is better, Claude Opus 4.8 or Claude Sonnet 4.6?
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 4.6 wins long documents, and costs 40% less.
Which is better for coding?
Claude Opus 4.8. It wins 8 of the 8 coding tests both models report; Claude Sonnet 4.6 wins none.
Which is cheaper?
Claude Opus 4.8 costs $5.00 per million input tokens and $25.00 per million output tokens; Claude Sonnet 4.6 costs $3.00 and $15.00. That makes Claude Sonnet 4.6 about 40% cheaper for the same work.
How do you compare the two?
We use the 69 benchmark tests both models have published scores on. The verdict counts the 30 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 39 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.