Claude Sonnet 4.5 vs Kimi K2 Instruct
Wins 7 of 7 areas
Coding · Agents · Reasoning · Facts · Math · Long documents · Following instructions
Wins 0 of 7 areas
—
Claude Sonnet 4.5 is the stronger all-rounder.Kimi K2 Instruct is cheaper.
Scores updated · 63 tests both models report · How we compare
Where each one wins
Tests won in each of the seven areas where both have results. 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 Sonnet 4.56 of 6 tests
- ReasoningHard problems that need careful thinking30Claude Sonnet 4.53 of 3 tests
- FactsGetting facts right instead of making them up30Claude Sonnet 4.53 of 3 tests
- AgentsCarrying out multi-step tasks on its own20Claude Sonnet 4.52 of 2 tests
- Following instructionsDoing exactly what it is asked20Claude Sonnet 4.52 of 2 tests
- MathCompetition and research-level math10Claude Sonnet 4.51 of 1 test
- Long documentsFinding answers in very long texts10Claude Sonnet 4.51 of 1 test
Math and long documents 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.
Kimi K2 Instruct costs 83% 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 Sonnet 4.5 pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+20.6points ahead
- Olympiad math problems with short answersIMOAnswerBench+20.1points ahead
- Fixes real GitHub issues in many programming languagesSWE-bench Multilingual+19.7points ahead
Where Kimi K2 Instruct pulls ahead
No clear win on a test scored out of 100.
Every test, side by side
All 63 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingClaude Sonnet 4.5
- SWE-bench MultilingualClaude Sonnet 4.5 by 19.76747.3+19.7
- SWE-bench ProClaude Sonnet 4.5 by 15.943.627.7+15.9
- SciCodeClaude Sonnet 4.5 by 1545.730.7+15
- Terminal-Bench HardClaude Sonnet 4.5 by 12.135.623.5+12.1
- SWE-bench VerifiedClaude Sonnet 4.5 by 11.477.265.8+11.4
- LiveCodeBench v6Claude Sonnet 4.5 by 10.36453.7+10.3
AgentsClaude Sonnet 4.5
- BrowseCompClaude Sonnet 4.5 by 1024.114.1+10
- GDPValClaude Sonnet 4.5 by 2.320.518.2+2.3
ReasoningClaude Sonnet 4.5
- Humanity's Last ExamClaude Sonnet 4.5 by 11.417.86.4+11.4
- GPQA DiamondClaude Sonnet 4.5 by 6.783.476.7+6.7
- CritPtClaude Sonnet 4.5 by 1.11.10+1.1
FactsClaude Sonnet 4.5
- AA-Omniscience · Non-hallucinationClaude Sonnet 4.5 by 20.650.930.3+20.6
- AA-Omniscience · AccuracyClaude Sonnet 4.5 by 7.532.925.4+7.5
- Vectara HHEM hallucination ratelower is betterClaude Sonnet 4.5 by 5.91217.9+5.9
Long documentsClaude Sonnet 4.5
- AA-LCRClaude Sonnet 4.5 by 19.372.353+19.3
Following instructionsClaude Sonnet 4.5
- IFBenchClaude Sonnet 4.5 by 15.657.341.7+15.6
- Multi-ChallengeClaude Sonnet 4.5 by 1.255.354.1+1.2
Other results45 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.
- AIME 2025Claude Sonnet 4.5 by 37.58749.5+37.5
- AIME25 no toolsClaude Sonnet 4.5 by 378851+37
- HMMT 2025Claude Sonnet 4.5 by 35.874.638.8+35.8
- HMMT25 no toolsClaude Sonnet 4.5 by 35.874.638.8+35.8
- FinSearchComp-T3 (Tools-allowed)Claude Sonnet 4.5 by 33.64410.4+33.6
- FinSearchComp-T3 w/ toolsClaude Sonnet 4.5 by 33.64410.4+33.6
- τ²-BenchClaude Sonnet 4.5 by 32.29865.8+32.2
- FinSearchComp-globalClaude Sonnet 4.5 by 31.360.829.5+31.3
- Seal-0 w/ toolsClaude Sonnet 4.5 by 28.253.425.2+28.2
- Frames w/ toolsClaude Sonnet 4.5 by 26.98558.1+26.9
- AA-OmniscienceClaude Sonnet 4.5 by 26.5-0.1-26.6+26.5
- Artificial Analysis Coding IndexClaude Sonnet 4.5 by 26.252.125.9+26.2
- AA Agentic IndexKimi K2 Instruct by 20.217.537.7+20.2
- BrowseComp-ZH w/ toolsClaude Sonnet 4.5 by 20.242.422.2+20.2
- Terminal-BenchClaude Sonnet 4.5 by 205030+20
- LiveCodeBenchClaude Sonnet 4.5 by 17.37153.7+17.3
- Longform Writing eval (Kimi K2 Thinking system card)Claude Sonnet 4.5 by 1779.862.8+17
- BrowseComp w/ toolsClaude Sonnet 4.5 by 16.724.17.4+16.7
- AIR-Bench 2024Claude Sonnet 4.5 by 15.789.874.1+15.7
- τ²-Bench (Retail)Claude Sonnet 4.5 by 15.686.270.6+15.6
- BrowseComp-ZHClaude Sonnet 4.5 by 13.642.428.8+13.6
- GAIA (text only)Claude Sonnet 4.5 by 1171.260.2+11
- theagentcompanyClaude Sonnet 4.5 by 114130+11
- Multi-SWE-BenchClaude Sonnet 4.5 by 10.844.333.5+10.8
- GPQA (unspecified)Claude Sonnet 4.5 by 9.283.474.2+9.2
- LiveCodeBenchV6 no toolsClaude Sonnet 4.5 by 7.96456.1+7.9
- ArtifactsBenchClaude Sonnet 4.5 by 7.361.554.2+7.3
- vectara_avg_summary_lengthClaude Sonnet 4.5 by 68.6 rating points127.859.2+68.6 rating
- Terminal-Bench w/ simulated tools (JSON)Claude Sonnet 4.5 by 6.55144.5+6.5
- MMLU-ProClaude Sonnet 4.5 by 6.487.581.1+6.4
- vectara_factual_consistencyClaude Sonnet 4.5 by 5.98882.1+5.9
- AA IntelligenceClaude Sonnet 4.5 by 5.420.715.3+5.4
- HarmBenchKimi K2 Instruct by 5.49297.4+5.4
- HLE (with tools)Claude Sonnet 4.5 by 5.13226.9+5.1
- xbench-DeepSearchClaude Sonnet 4.5 by 56661+5
- OJ-Bench (cpp)Claude Sonnet 4.5 by 4.930.425.5+4.9
- OJ-Bench (cpp) no toolsClaude Sonnet 4.5 by 4.930.425.5+4.9
- τ²-Bench Telecom (AA run)Claude Sonnet 4.5 by 4.778.173.4+4.7
- vectara_answer_rateKimi K2 Instruct by 395.698.6+3
- MMLU-ReduxClaude Sonnet 4.5 by 2.995.692.7+2.9
- XSTestKimi K2 Instruct by 1.896.498.2+1.8
- anthropic_red_teamKimi K2 Instruct by 1.198.299.3+1.1
- HealthBenchtie44.243.8tie
- HealthBench no toolstie44.243.8tie
- simple_safety_teststie100100tie
Questions people ask
Which is better, Claude Sonnet 4.5 or Kimi K2 Instruct?
Claude Sonnet 4.5 wins all seven areas where both have results: coding, agents, reasoning, facts, math, long documents and following instructions. Kimi K2 Instruct wins none, but costs 83% less.
Which is better for coding?
Claude Sonnet 4.5. It wins 6 of the 6 coding tests both models report; Kimi K2 Instruct wins none.
Which is cheaper?
Claude Sonnet 4.5 costs $3.00 per million input tokens and $15.00 per million output tokens; Kimi K2 Instruct costs $0.60 and $2.50. That makes Kimi K2 Instruct about 83% cheaper for the same work.
How do you compare the two?
We use the 63 benchmark tests both models have published scores on. The verdict counts the 18 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 45 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.