Claude Opus 4.8 vs Qwen3.5 397B A17B
Wins 6 of 8 areas
Coding · Agents · Reasoning · Facts · Images and charts · Math
Wins 1 of 8 areas
Following instructions
Claude Opus 4.8 is the stronger all-rounder.Qwen3.5 397B A17B is cheaper and better at following instructions.
Scores updated · 48 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 software70Claude Opus 4.87 of 7 tests
- AgentsCarrying out multi-step tasks on its own50Claude Opus 4.85 of 5 tests
- MathCompetition and research-level math50Claude Opus 4.85 of 5 tests
- ReasoningHard problems that need careful thinking30Claude Opus 4.83 of 3 tests
- FactsGetting facts right instead of making them up30Claude Opus 4.83 of 3 tests
- Images and chartsUnderstanding pictures, charts and video10Claude Opus 4.81 of 1 test
- Following instructionsDoing exactly what it is asked01Qwen3.5 397B A17B1 of 1 test
- Long documentsFinding answers in very long texts00Even0 each · 1 tie
Images and charts, 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.
Qwen3.5 397B A17B costs 86% 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+60.4points ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+49.6points ahead
- Unpublished advanced math problemsFrontierMath Tiers 1-3 (v2)+48.8points ahead
Where Qwen3.5 397B A17B pulls ahead
- Follows unfamiliar, precisely checkable instructionsIFBench+16.6points ahead
Every test, side by side
All 48 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 33.384.651.3+33.3
- SWE-bench ProClaude Opus 4.8 by 18.369.250.9+18.3
- Terminal-Bench HardClaude Opus 4.8 by 17.458.340.9+17.4
- LMArena · WebDevClaude Opus 4.8 by 157 rating points15561399+157 rating
- SWE-bench MultilingualClaude Opus 4.8 by 15.184.469.3+15.1
- SWE-bench VerifiedClaude Opus 4.8 by 12.288.676.4+12.2
- SciCodeClaude Opus 4.8 by 9.654.444.8+9.6
AgentsClaude Opus 4.8
- GDPValClaude Opus 4.8 by 3347.814.8+33
- AA ApexAgentsClaude Opus 4.8 by 24.139.415.3+24.1
- Terminal-Bench 4.0Claude Opus 4.8 by 21.721.70+21.7
- τ-Bench V3 · BankingClaude Opus 4.8 by 20.834.213.4+20.8
- BrowseCompClaude Opus 4.8 by 15.384.369+15.3
ReasoningClaude Opus 4.8
- Humanity's Last ExamClaude Opus 4.8 by 19.748.729+19.7
- CritPtClaude Opus 4.8 by 19.220.91.7+19.2
- GPQA DiamondClaude Opus 4.8 by 2.79289.3+2.7
FactsClaude Opus 4.8
- AA-Omniscience · Non-hallucinationClaude Opus 4.8 by 49.660.711.1+49.6
- SimpleQA VerifiedClaude Opus 4.8 by 275326+27
- AA-Omniscience · AccuracyClaude Opus 4.8 by 1848.830.8+18
Images and chartsClaude Opus 4.8
- LMArena · VisionClaude Opus 4.8 by 31 rating points12951264+31 rating
MathClaude Opus 4.8
- USAMO 2026Claude Opus 4.8 by 60.496.736.3+60.4
- FrontierMath Tiers 1-3 (v2)Claude Opus 4.8 by 48.88031.2+48.8
- AIME 2026Claude Opus 4.8 by 8.710091.3+8.7
- HMMT Feb. 2026Claude Opus 4.8 by 7.695.587.9+7.6
- IMOAnswerBenchClaude Opus 4.8 by 2.683.580.9+2.6
Following instructionsQwen3.5 397B A17B
- IFBenchQwen3.5 397B A17B by 16.662.278.8+16.6
Other results22 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.
- GDPval-AA v2Claude Opus 4.8 by 631 rating points1593962+631 rating
- AA-OmniscienceClaude Opus 4.8 by 59.528.8-30.8+59.5
- Toolathlon VerifiedClaude Opus 4.8 by 39.279.940.7+39.2
- NL2RepoClaude Opus 4.8 by 37.569.732.2+37.5
- AA Agentic IndexClaude Opus 4.8 by 3242.610.6+32
- Artificial Analysis Coding IndexClaude Opus 4.8 by 26.174.348.2+26.1
- AA IntelligenceClaude Opus 4.8 by 23.441.818.4+23.4
- Terminal-Bench 2.0Claude Opus 4.8 by 22.174.652.5+22.1
- ToolathlonClaude Opus 4.8 by 21.659.938.3+21.6
- AgentWorldBench - MCPQwen3.5 397B A17B by 13.454.968.3+13.4
- HLE (with tools)Claude Opus 4.8 by 9.657.948.3+9.6
- AgentWorldBench - AndroidClaude Opus 4.8 by 6.661.554.9+6.6
- AgentWorldBench - WebClaude Opus 4.8 by 6.154.748.5+6.1
- AgentWorldBench - OSClaude Opus 4.8 by 5.766.660.9+5.7
- AgentWorldBench - SearchClaude Opus 4.8 by 4.335.130.8+4.3
- AgentWorldBench - TerminalClaude Opus 4.8 by 3.959.255.3+3.9
- HMMT Nov. 2025Claude Opus 4.8 by 3.896.592.7+3.8
- IncludeClaude Opus 4.8 by 287.685.6+2
- AgentWorldBench - OverallClaude Opus 4.8 by 1.956.654.7+1.9
- τ²-Bench Telecom (AA run)Qwen3.5 397B A17B by 1.294.495.6+1.2
- WideSearchQwen3.5 397B A17B by 1.172.974+1.1
- AgentWorldBench - SWEtie64.164.4tie
Questions people ask
Which is better, Claude Opus 4.8 or Qwen3.5 397B A17B?
Claude Opus 4.8 wins six of the eight areas we test: coding, agents, reasoning, facts, images and charts and math. Qwen3.5 397B A17B wins following instructions, and costs 86% less. They are level on long documents.
Which is better for coding?
Claude Opus 4.8. It wins 7 of the 7 coding tests both models report; Qwen3.5 397B A17B wins none.
Which is cheaper?
Claude Opus 4.8 costs $5.00 per million input tokens and $25.00 per million output tokens; Qwen3.5 397B A17B costs $0.60 and $3.60. That makes Qwen3.5 397B A17B about 86% cheaper for the same work.
How do you compare the two?
We use the 48 benchmark tests both models have published scores on. The verdict counts the 26 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 22 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.