Kimi K2.5 vs Kimi K2 Instruct
Wins 6 of 7 areas
Coding · Reasoning · Facts · Math · Long documents · Following instructions
Wins 0 of 7 areas
—
Kimi K2.5 is the stronger all-rounder.Kimi K2 Instruct is cheaper.
Scores updated · 35 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 software60Kimi K2.56 of 6 tests
- ReasoningHard problems that need careful thinking30Kimi K2.53 of 3 tests
- FactsGetting facts right instead of making them up30Kimi K2.53 of 3 tests
- Following instructionsDoing exactly what it is asked20Kimi K2.52 of 2 tests
- MathCompetition and research-level math10Kimi K2.51 of 1 test
- Long documentsFinding answers in very long texts10Kimi K2.51 of 1 test
- AgentsCarrying out multi-step tasks on its own11Even1 each
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 14% 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
- Finds hard-to-locate facts by browsing the webBrowseComp+60.8points ahead
- Olympiad math problems with short answersIMOAnswerBench+36points ahead
- Recent programming contest problemsLiveCodeBench v6+31.3points ahead
Where Kimi K2 Instruct pulls ahead
- Real work tasks from 44 professionsGDPVal+1.2points ahead
Every test, side by side
All 35 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingKimi K2.5
- LiveCodeBench v6Kimi K2.5 by 31.38553.7+31.3
- SWE-bench MultilingualKimi K2.5 by 25.77347.3+25.7
- SWE-bench ProKimi K2.5 by 2350.727.7+23
- SciCodeKimi K2.5 by 18.34930.7+18.3
- Terminal-Bench HardKimi K2.5 by 11.334.823.5+11.3
- SWE-bench VerifiedKimi K2.5 by 1176.865.8+11
AgentsEven
- BrowseCompKimi K2.5 by 60.874.914.1+60.8
- GDPValKimi K2 Instruct by 1.21718.2+1.2
ReasoningKimi K2.5
- Humanity's Last ExamKimi K2.5 by 24.330.76.4+24.3
- GPQA DiamondKimi K2.5 by 11.287.976.7+11.2
- CritPtKimi K2.5 by 3.13.10+3.1
FactsKimi K2.5
- AA-Omniscience · AccuracyKimi K2.5 by 9.835.225.4+9.8
- AA-Omniscience · Non-hallucinationKimi K2.5 by 434.330.3+4
- Vectara HHEM hallucination ratelower is betterKimi K2.5 by 3.714.217.9+3.7
Following instructionsKimi K2.5
- IFBenchKimi K2.5 by 28.570.241.7+28.5
- Multi-ChallengeKimi K2.5 by 7.361.454.1+7.3
Other results17 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.
- HMMT 2025Kimi K2.5 by 56.695.438.8+56.6
- AIME 2025Kimi K2.5 by 46.696.149.5+46.6
- BrowseComp-ZHKimi K2.5 by 33.562.328.8+33.5
- OJ-Bench (cpp)Kimi K2.5 by 31.957.425.5+31.9
- HLE (with tools)Kimi K2.5 by 23.350.226.9+23.3
- τ²-Bench Telecom (AA run)Kimi K2.5 by 22.595.973.4+22.5
- Artificial Analysis Coding IndexKimi K2.5 by 20.946.825.9+20.9
- τ²-BenchKimi K2.5 by 19.685.465.8+19.6
- AA-OmniscienceKimi K2.5 by 19.3-7.3-26.6+19.3
- AA Agentic IndexKimi K2 Instruct by 1621.737.7+16
- xbench-DeepSearchKimi K2.5 by 15.776.761+15.7
- AA IntelligenceKimi K2.5 by 8.223.515.3+8.2
- LiveBenchKimi K2 Instruct by 7.369.176.4+7.3
- vectara_answer_rateKimi K2 Instruct by 6.492.298.6+6.4
- MMLU-ProKimi K2.5 by 687.181.1+6
- vectara_avg_summary_lengthKimi K2.5 by 52.8 rating points11259.2+52.8 rating
- vectara_factual_consistencyKimi K2.5 by 3.785.882.1+3.7
Questions people ask
Which is better, Kimi K2.5 or Kimi K2 Instruct?
Kimi K2.5 wins six of the seven areas where both have results: coding, reasoning, facts, math, long documents and following instructions. Kimi K2 Instruct wins none, but costs 14% less. They are level on agents.
Which is better for coding?
Kimi K2.5. It wins 6 of the 6 coding tests both models report; Kimi K2 Instruct wins none.
Which is cheaper?
Kimi K2.5 costs $0.60 per million input tokens and $3.00 per million output tokens; Kimi K2 Instruct costs $0.60 and $2.50. That makes Kimi K2 Instruct about 14% cheaper for the same work.
How do you compare the two?
We use the 35 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 17 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.