GLM-4.6 vs Kimi K2.5
Wins 0 of 7 areas
—
Wins 7 of 7 areas
Coding · Agents · Reasoning · Facts · Math · Long documents · Following instructions
Kimi K2.5 is the stronger all-rounder.GLM-4.6 is cheaper.
Scores updated · 41 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 software17Kimi K2.57 of 8 tests
- ReasoningHard problems that need careful thinking03Kimi K2.53 of 3 tests
- AgentsCarrying out multi-step tasks on its own02Kimi K2.52 of 3 tests · 1 tie
- Following instructionsDoing exactly what it is asked02Kimi K2.52 of 2 tests
- FactsGetting facts right instead of making them up12Kimi K2.52 of 3 tests
- MathCompetition and research-level math01Kimi K2.51 of 1 test
- Long documentsFinding answers in very long texts01Kimi K2.51 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.
GLM-4.6 costs 23% 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 Kimi K2.5 pulls ahead
- Long, multi-file coding tasks in real codebasesSWE-bench Pro+41points ahead
- Finds hard-to-locate facts by browsing the webBrowseComp+29.8points ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+28.4points ahead
Where GLM-4.6 pulls ahead
- Command-line tasks in a real terminalTerminal-Bench 2.1+3.7points ahead
Every test, side by side
All 41 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingKimi K2.5
- SWE-bench ProKimi K2.5 by 419.750.7+41
- SWE-bench MultilingualKimi K2.5 by 19.253.873+19.2
- SciCodeKimi K2.5 by 10.638.449+10.6
- Terminal-Bench HardKimi K2.5 by 9.82534.8+9.8
- LMArena · WebDevKimi K2.5 by 95 rating points13411436+95 rating
- SWE-bench VerifiedKimi K2.5 by 8.86876.8+8.8
- Terminal-Bench 2.1GLM-4.6 by 3.749.445.7+3.7
- LiveCodeBench v6Kimi K2.5 by 2.282.885+2.2
AgentsKimi K2.5
- BrowseCompKimi K2.5 by 29.845.174.9+29.8
- GDPValKimi K2.5 by 4.21317.2+4.2
- τ-Bench V3 · Bankingtie13.414.2tie
ReasoningKimi K2.5
- Humanity's Last ExamKimi K2.5 by 16.214.530.7+16.2
- GPQA DiamondKimi K2.5 by 9.97887.9+9.9
- CritPtKimi K2.5 by 21.13.1+2
FactsKimi K2.5
- AA-Omniscience · Non-hallucinationKimi K2.5 by 28.45.934.3+28.4
- AA-Omniscience · AccuracyKimi K2.5 by 8.326.935.2+8.3
- Vectara HHEM hallucination ratelower is betterGLM-4.6 by 4.79.514.2+4.7
Following instructionsKimi K2.5
- IFBenchKimi K2.5 by 26.843.470.2+26.8
- Multi-ChallengeKimi K2.5 by 6.554.961.4+6.5
Other results20 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-OmniscienceKimi K2.5 by 34.6-41.9-7.3+34.6
- Terminal-Bench 2.0Kimi K2.5 by 26.324.550.8+26.3
- τ²-Bench Telecom (AA run)Kimi K2.5 by 25.470.595.9+25.4
- HLE (with tools)Kimi K2.5 by 19.830.450.2+19.8
- browsecomp_with_context_managerKimi K2.5 by 17.457.574.9+17.4
- BrowseComp-ZHKimi K2.5 by 12.849.562.3+12.8
- τ²-BenchKimi K2.5 by 10.275.285.4+10.2
- HMMT 2025Kimi K2.5 by 6.788.795.4+6.7
- xbench-DeepSearchKimi K2.5 by 6.77076.7+6.7
- HMMT Feb. 2025Kimi K2.5 by 6.289.295.4+6.2
- AA IntelligenceKimi K2.5 by 518.523.5+5
- vectara_factual_consistencyGLM-4.6 by 4.790.585.8+4.7
- MMLU-ProKimi K2.5 by 3.983.287.1+3.9
- vectara_avg_summary_lengthKimi K2.5 by 34.8 rating points77.2112+34.8 rating
- AA Agentic IndexKimi K2.5 by 3.118.621.7+3.1
- HMMT Nov. 2025GLM-4.6 by 2.591.789.2+2.5
- vectara_answer_rateGLM-4.6 by 2.394.592.2+2.3
- AIME 2025Kimi K2.5 by 2.293.996.1+2.2
- frontiermath_tier_4_v1Kimi K2.5 by 2.12.14.2+2.1
- Artificial Analysis Coding IndexKimi K2.5 by 145.846.8+1
Questions people ask
Which is better, GLM-4.6 or Kimi K2.5?
Kimi K2.5 wins all seven areas where both have results: coding, agents, reasoning, facts, math, long documents and following instructions. GLM-4.6 wins none, but costs 23% less.
Which is better for coding?
Kimi K2.5. It wins 7 of the 8 coding tests both models report; GLM-4.6 wins 1.
Which is cheaper?
GLM-4.6 costs $0.57 per million input tokens and $2.20 per million output tokens; Kimi K2.5 costs $0.60 and $3.00. That makes GLM-4.6 about 23% cheaper for the same work.
How do you compare the two?
We use the 41 benchmark tests both models have published scores on. The verdict counts the 21 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 20 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.