Claude Opus 4.5 vs Kimi K2.5
Wins 4 of 8 areas
Coding · Reasoning · Facts · Long documents
Wins 2 of 8 areas
Images and charts · Following instructions
Claude Opus 4.5 wins more areas, narrowly.Kimi K2.5 is cheaper and better at images and charts.
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.
- FactsGetting facts right instead of making them up40Claude Opus 4.54 of 4 tests
- CodingWriting and fixing software41Claude Opus 4.54 of 7 tests · 2 ties
- ReasoningHard problems that need careful thinking21Claude Opus 4.52 of 4 tests · 1 tie
- Long documentsFinding answers in very long texts10Claude Opus 4.51 of 2 tests · 1 tie
- Images and chartsUnderstanding pictures, charts and video04Kimi K2.54 of 4 tests
- Following instructionsDoing exactly what it is asked02Kimi K2.52 of 2 tests
- AgentsCarrying out multi-step tasks on its own22Even2 each
- MathCompetition and research-level math11Even1 each · 1 tie
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.
Kimi K2.5 costs 88% 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 4.5 pulls ahead
- Real work tasks from 44 professionsGDPVal+30.1points ahead
- Abstract visual puzzles that people can solveARC-AGI-2+25.8points ahead
- Hard command-line tasks in a real terminalTerminal-Bench Hard+12.2points ahead
Where Kimi K2.5 pulls ahead
- Fixes real GitHub issues in many programming languagesSWE-bench Multilingual+66points ahead
- Finds hard-to-locate facts by browsing the webBrowseComp+37.9points ahead
- Follows unfamiliar, precisely checkable instructionsIFBench+12.2points 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 4.5
- SWE-bench MultilingualKimi K2.5 by 66773+66
- Terminal-Bench HardClaude Opus 4.5 by 12.24734.8+12.2
- SWE-bench VerifiedClaude Opus 4.5 by 4.180.976.8+4.1
- LMArena · WebDevClaude Opus 4.5 by 33 rating points14691436+33 rating
- SWE-bench ProClaude Opus 4.5 by 1.35250.7+1.3
- SciCodetie49.549tie
- LiveCodeBench v6tie84.885tie
AgentsEven
- BrowseCompKimi K2.5 by 37.93774.9+37.9
- GDPValClaude Opus 4.5 by 30.147.317.2+30.1
- OSWorld-VerifiedClaude Opus 4.5 by 366.363.3+3
- MCP AtlasKimi K2.5 by 1.562.363.8+1.5
ReasoningClaude Opus 4.5
- ARC-AGI-2Claude Opus 4.5 by 25.837.611.8+25.8
- CritPtClaude Opus 4.5 by 1.54.63.1+1.5
- GPQA DiamondKimi K2.5 by 1.386.687.9+1.3
- Humanity's Last Examtie30.130.7tie
FactsClaude Opus 4.5
- AA-Omniscience · AccuracyClaude Opus 4.5 by 11.446.635.2+11.4
- SimpleQA VerifiedClaude Opus 4.5 by 11.445.734.3+11.4
- AA-Omniscience · Non-hallucinationClaude Opus 4.5 by 4.73934.3+4.7
- Vectara HHEM hallucination ratelower is betterClaude Opus 4.5 by 3.310.914.2+3.3
Images and chartsKimi K2.5
- CharXiv (RQ)Kimi K2.5 by 10.367.277.5+10.3
- MathVistaKimi K2.5 by 9.980.290.1+9.9
- OCRBenchKimi K2.5 by 5.886.592.3+5.8
- MMMU-ProKimi K2.5 by 1.47475.4+1.4
MathEven
- IMOAnswerBenchClaude Opus 4.5 by 2.28481.8+2.2
- HMMT Feb. 2026Kimi K2.5 by 1.885.387.1+1.8
- AIME 2026tie95.195.8tie
Long documentsClaude Opus 4.5
- LongBench v2Claude Opus 4.5 by 3.464.461+3.4
- AA-LCRtie77.378tie
Following instructionsKimi K2.5
- IFBenchKimi K2.5 by 12.25870.2+12.2
- Multi-ChallengeKimi K2.5 by 2.45961.4+2.4
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.
- AA Agentic IndexClaude Opus 4.5 by 37.959.621.7+37.9
- AA-OmniscienceClaude Opus 4.5 by 21.314-7.3+21.3
- ARC-AGI-1Claude Opus 4.5 by 16.78265.3+16.7
- InfoVQA (val)Kimi K2.5 by 15.776.992.6+15.7
- ToolathlonClaude Opus 4.5 by 15.743.527.8+15.7
- τ²-BenchClaude Opus 4.5 by 12.898.285.4+12.8
- NL2RepoClaude Opus 4.5 by 11.243.232+11.2
- MotionBenchKimi K2.5 by 10.160.370.4+10.1
- Seal-0Kimi K2.5 by 9.747.757.4+9.7
- WorldVQAKimi K2.5 by 9.536.846.3+9.5
- PaperBenchClaude Opus 4.5 by 9.472.963.5+9.4
- CyberGymClaude Opus 4.5 by 9.350.641.3+9.3
- Terminal-Bench 2.0Claude Opus 4.5 by 8.559.350.8+8.5
- ZeroBenchKimi K2.5 by 8311+8
- MATH-VisionKimi K2.5 by 7.177.184.2+7.1
- HLE (with tools)Kimi K2.5 by 743.250.2+7
- LiveBenchClaude Opus 4.5 by 6.97669.1+6.9
- vectara_answer_rateClaude Opus 4.5 by 6.598.792.2+6.5
- τ²-Bench Telecom (AA run)Kimi K2.5 by 6.489.595.9+6.4
- AA IntelligenceClaude Opus 4.5 by 5.629.123.5+5.6
- HMMT Nov. 2025Claude Opus 4.5 by 4.193.389.2+4.1
- swe_bench_bashClaude Opus 4.5 by 3.674.470.8+3.6
- WideSearch (item-f1)Claude Opus 4.5 by 3.576.272.7+3.5
- AIME 2025Kimi K2.5 by 3.392.896.1+3.3
- vectara_factual_consistencyClaude Opus 4.5 by 3.389.185.8+3.3
- MMVUKimi K2.5 by 3.177.380.4+3.1
- OJ-Bench (cpp)Kimi K2.5 by 2.854.657.4+2.8
- HMMT Feb. 2025Kimi K2.5 by 2.592.995.4+2.5
- MMLU-ProClaude Opus 4.5 by 2.489.587.1+2.4
- VideoMMMUKimi K2.5 by 2.284.486.6+2.2
- ZeroBench (w/ tools)Kimi K2.5 by 2911+2
- FinSearchCompT2&T3Kimi K2.5 by 1.666.267.8+1.6
- SimpleVQAKimi K2.5 by 1.569.771.2+1.5
- OmniDocBench 1.5Kimi K2.5 by 1.187.788.8+1.1
- Artificial Analysis Coding IndexClaude Opus 4.5 by 147.846.8+1
- DeepSearchQA (F1)Kimi K2.5 by 176.177.1+1
- AIME 2026 I (Tools-allowed)tie93.392.5tie
- vectara_avg_summary_lengthtie114.5112tie
- frontiermath_tier_4_v1tie4.24.2tie
Questions people ask
Which is better, Claude Opus 4.5 or Kimi K2.5?
Claude Opus 4.5 wins four of the eight areas we test: coding, reasoning, facts and long documents. Kimi K2.5 wins images and charts and following instructions, and costs 88% less. They are level on agents and math.
Which is better for coding?
Claude Opus 4.5. It wins 4 of the 7 coding tests both models report; Kimi K2.5 wins 1, and 2 are ties.
Which is cheaper?
Claude Opus 4.5 costs $5.00 per million input tokens and $25.00 per million output tokens; Kimi K2.5 costs $0.60 and $3.00. That makes Kimi K2.5 about 88% 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.