Kimi K2.5 vs Qwen3.6 27B
Wins 5 of 8 areas
Coding · Reasoning · Images and charts · Math · Following instructions
Wins 1 of 8 areas
Agents
Kimi K2.5 is the stronger all-rounder.Qwen3.6 27B is better at agents.
Both rank among the ten best models we track in images and charts.
Scores updated · 36 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.
- ReasoningHard problems that need careful thinking30Kimi K2.53 of 3 tests
- MathCompetition and research-level math30Kimi K2.53 of 3 tests
- Images and chartsUnderstanding pictures, charts and video20Kimi K2.52 of 4 tests · 2 ties
- CodingWriting and fixing software32Kimi K2.53 of 7 tests · 2 ties
- Following instructionsDoing exactly what it is asked10Kimi K2.51 of 1 test
- AgentsCarrying out multi-step tasks on its own02Qwen3.6 27B2 of 2 tests
- FactsGetting facts right instead of making them up11Even1 each
- Long documentsFinding answers in very long texts00Even0 each · 1 tie
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.
Kimi K2.5 costs 14% 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 K2.5 pulls ahead
- Answers hard knowledge questions correctlyAA-Omniscience · Accuracy+15.6points ahead
- Very hard expert questions across many subjectsHumanity's Last Exam+7.6points ahead
- Code for real scientific research problemsSciCode+6.2points ahead
Where Qwen3.6 27B pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+16.4points ahead
- Command-line tasks in a real terminalTerminal-Bench 2.1+15points ahead
- Real work tasks from 44 professionsGDPVal+7.2points ahead
Every test, side by side
All 36 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingKimi K2.5
- Terminal-Bench 2.1Qwen3.6 27B by 1545.760.7+15
- SciCodeKimi K2.5 by 6.24942.8+6.2
- SWE-bench ProQwen3.6 27B by 2.850.753.5+2.8
- SWE-bench MultilingualKimi K2.5 by 1.77371.3+1.7
- LiveCodeBench v6Kimi K2.5 by 1.18583.9+1.1
- SWE-bench Verifiedtie76.877.2tie
- Terminal-Bench Hardtie34.834.8tie
AgentsQwen3.6 27B
- GDPValQwen3.6 27B by 7.217.224.4+7.2
- τ-Bench V3 · BankingQwen3.6 27B by 2.514.216.7+2.5
ReasoningKimi K2.5
- Humanity's Last ExamKimi K2.5 by 7.630.723.1+7.6
- GPQA DiamondKimi K2.5 by 3.787.984.2+3.7
- CritPtKimi K2.5 by 23.11.1+2
FactsEven
- AA-Omniscience · Non-hallucinationQwen3.6 27B by 16.434.350.7+16.4
- AA-Omniscience · AccuracyKimi K2.5 by 15.635.219.6+15.6
Images and chartsKimi K2.5
- OCRBenchKimi K2.5 by 2.992.389.4+2.9
- MathVistaKimi K2.5 by 2.790.187.4+2.7
- CharXiv (RQ)tie77.578.4tie
- MMMU-Protie75.474.6tie
MathKimi K2.5
- HMMT Feb. 2026Kimi K2.5 by 2.887.184.3+2.8
- AIME 2026Kimi K2.5 by 1.795.894.1+1.7
- IMOAnswerBenchKimi K2.5 by 181.880.8+1
Following instructionsKimi K2.5
- IFBenchKimi K2.5 by 2.670.267.6+2.6
Other results13 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.
- SimpleVQAKimi K2.5 by 15.171.256.1+15.1
- AA-OmniscienceKimi K2.5 by 12.7-7.3-20+12.7
- Terminal-Bench 2.0Qwen3.6 27B by 8.550.859.3+8.5
- Artificial Analysis Coding IndexQwen3.6 27B by 6.946.853.7+6.9
- HMMT 2025Kimi K2.5 by 4.795.490.7+4.7
- NL2RepoQwen3.6 27B by 4.23236.2+4.2
- VideoMMMUKimi K2.5 by 2.286.684.4+2.2
- AA IntelligenceKimi K2.5 by 2.123.521.4+2.1
- τ²-Bench Telecom (AA run)Kimi K2.5 by 1.795.994.2+1.7
- AA Agentic IndexKimi K2.5 by 1.621.720.1+1.6
- HMMT Feb. 2025Kimi K2.5 by 1.695.493.8+1.6
- HMMT Nov. 2025Qwen3.6 27B by 1.589.290.7+1.5
- MMLU-Protie87.186.2tie
Questions people ask
Which is better, Kimi K2.5 or Qwen3.6 27B?
Kimi K2.5 wins five of the eight areas we test: coding, reasoning, images and charts, math and following instructions. Qwen3.6 27B wins agents. They are level on facts and long documents.
Which is better for coding?
Kimi K2.5. It wins 3 of the 7 coding tests both models report; Qwen3.6 27B wins 2, and 2 are ties.
Which is cheaper?
Kimi K2.5 costs $0.60 per million input tokens and $3.00 per million output tokens; Qwen3.6 27B costs $0.60 and $3.60. That makes Kimi K2.5 about 14% cheaper for the same work.
How do you compare the two?
We use the 36 benchmark tests both models have published scores on. The verdict counts the 23 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 13 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.