Claude Sonnet 4.5 vs Kimi K2 Thinking
Wins 3 of 7 areas
Coding · Facts · Long documents
Wins 4 of 7 areas
Agents · Reasoning · Math · Following instructions
Kimi K2 Thinking wins more areas, narrowly.Claude Sonnet 4.5 is better at coding.
Scores updated · 49 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.
- AgentsCarrying out multi-step tasks on its own02Kimi K2 Thinking2 of 2 tests
- ReasoningHard problems that need careful thinking12Kimi K2 Thinking2 of 4 tests · 1 tie
- Following instructionsDoing exactly what it is asked01Kimi K2 Thinking1 of 2 tests · 1 tie
- MathCompetition and research-level math01Kimi K2 Thinking1 of 1 test
- CodingWriting and fixing software41Claude Sonnet 4.54 of 5 tests
- FactsGetting facts right instead of making them up20Claude Sonnet 4.52 of 2 tests
- Long documentsFinding answers in very long texts10Claude Sonnet 4.51 of 2 tests · 1 tie
Math 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.
Kimi K2 Thinking costs 83% less for the same work.
The biggest differences
The three tests each model wins by the widest margin. Scores are out of 100.
Where Kimi K2 Thinking pulls ahead
- Recent programming contest problemsLiveCodeBench v6+19.1points ahead
- Finds hard-to-locate facts by browsing the webBrowseComp+17.4points ahead
- Olympiad math problems with short answersIMOAnswerBench+12.7points ahead
Where Claude Sonnet 4.5 pulls ahead
- Common-sense trick questionsSimpleBench+28points ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+26.5points ahead
- Questions about very long textsLongBench v2+16.7points ahead
Every test, side by side
All 49 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingClaude Sonnet 4.5
- LiveCodeBench v6Kimi K2 Thinking by 19.16483.1+19.1
- SWE-bench VerifiedClaude Sonnet 4.5 by 13.877.263.4+13.8
- SWE-bench MultilingualClaude Sonnet 4.5 by 5.96761.1+5.9
- Terminal-Bench HardClaude Sonnet 4.5 by 4.535.631.1+4.5
- SciCodeClaude Sonnet 4.5 by 3.345.742.4+3.3
AgentsKimi K2 Thinking
- BrowseCompKimi K2 Thinking by 17.424.141.5+17.4
- GDPValKimi K2 Thinking by 420.524.5+4
ReasoningKimi K2 Thinking
- SimpleBenchClaude Sonnet 4.5 by 2854.326.3+28
- Humanity's Last ExamKimi K2 Thinking by 617.823.8+6
- CritPtKimi K2 Thinking by 1.51.12.6+1.5
- GPQA Diamondtie83.483.8tie
FactsClaude Sonnet 4.5
- AA-Omniscience · Non-hallucinationClaude Sonnet 4.5 by 26.550.924.4+26.5
- AA-Omniscience · AccuracyClaude Sonnet 4.5 by 232.930.9+2
Long documentsClaude Sonnet 4.5
- LongBench v2Claude Sonnet 4.5 by 16.761.845.1+16.7
- AA-LCRtie72.372tie
Following instructionsKimi K2 Thinking
- IFBenchKimi K2 Thinking by 10.857.368.1+10.8
- Multi-Challengetie55.355.4tie
Other results31 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 IndexKimi K2 Thinking by 30.417.547.9+30.4
- HMMT Feb. 2025Kimi K2 Thinking by 25.867.593.3+25.8
- τ²-BenchClaude Sonnet 4.5 by 23.79874.3+23.7
- AA-OmniscienceClaude Sonnet 4.5 by 21.3-0.1-21.4+21.3
- BrowseComp-ZHKimi K2 Thinking by 19.942.462.3+19.9
- LiveCodeBenchV6 no toolsKimi K2 Thinking by 19.16483.1+19.1
- OJ-Bench (cpp)Kimi K2 Thinking by 18.330.448.7+18.3
- OJ-Bench (cpp) no toolsKimi K2 Thinking by 18.330.448.7+18.3
- Artificial Analysis Coding IndexClaude Sonnet 4.5 by 17.352.134.8+17.3
- τ²-Bench Telecom (AA run)Kimi K2 Thinking by 14.978.193+14.9
- HMMT 2025Kimi K2 Thinking by 14.874.689.4+14.8
- HMMT25 no toolsKimi K2 Thinking by 14.874.689.4+14.8
- Terminal-Bench 2.0Claude Sonnet 4.5 by 14.35035.7+14.3
- HealthBenchKimi K2 Thinking by 13.844.258+13.8
- HealthBench no toolsKimi K2 Thinking by 13.844.258+13.8
- HLE (with tools)Kimi K2 Thinking by 12.93244.9+12.9
- MRCRClaude Sonnet 4.5 by 11.255.444.2+11.2
- xbench-DeepSearchKimi K2 Thinking by 106676+10
- Arena-Hard (Hard Prompt)Kimi K2 Thinking by 8.663.371.9+8.6
- LiveCodeBenchKimi K2 Thinking by 8.27179.2+8.2
- swe_bench_bashClaude Sonnet 4.5 by 871.463.4+8
- AIME 2025Kimi K2 Thinking by 7.58794.5+7.5
- HMMT Nov. 2025Kimi K2 Thinking by 7.581.789.2+7.5
- AIME25 no toolsKimi K2 Thinking by 6.58894.5+6.5
- Longform Writing eval (Kimi K2 Thinking system card)Claude Sonnet 4.5 by 679.873.8+6
- frontiermath_tier_4_v1Claude Sonnet 4.5 by 4.24.20+4.2
- Arena-Hard (Creative Writing)Kimi K2 Thinking by 3.476.780.1+3.4
- FinSearchComp-T3 (Tools-allowed)Kimi K2 Thinking by 3.44447.4+3.4
- MMLU-ProClaude Sonnet 4.5 by 2.987.584.6+2.9
- AA IntelligenceKimi K2 Thinking by 1.320.722+1.3
- MMLU-ReduxClaude Sonnet 4.5 by 1.295.694.4+1.2
Questions people ask
Which is better, Claude Sonnet 4.5 or Kimi K2 Thinking?
Kimi K2 Thinking wins four of the seven areas where both have results: agents, reasoning, math and following instructions. Claude Sonnet 4.5 wins coding, facts and long documents.
Which is better for coding?
Claude Sonnet 4.5. It wins 4 of the 5 coding tests both models report; Kimi K2 Thinking wins 1.
Which is cheaper?
Claude Sonnet 4.5 costs $3.00 per million input tokens and $15.00 per million output tokens; Kimi K2 Thinking costs $0.60 and $2.50. That makes Kimi K2 Thinking about 83% cheaper for the same work.
How do you compare the two?
We use the 49 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 31 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.