Gemma 4 26B A4B vs Kimi K2 Instruct
Wins 4 of 7 areas
Reasoning · Math · Long documents · Following instructions
Wins 2 of 7 areas
Coding · Facts
Gemma 4 26B A4B wins more areas, narrowly.Kimi K2 Instruct is better at coding.
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.
- ReasoningHard problems that need careful thinking20Gemma 4 26B A4B2 of 3 tests · 1 tie
- MathCompetition and research-level math10Gemma 4 26B A4B1 of 1 test
- Long documentsFinding answers in very long texts10Gemma 4 26B A4B1 of 1 test
- Following instructionsDoing exactly what it is asked10Gemma 4 26B A4B1 of 1 test
- CodingWriting and fixing software24Kimi K2 Instruct4 of 6 tests
- FactsGetting facts right instead of making them up12Kimi K2 Instruct2 of 3 tests
- AgentsCarrying out multi-step tasks on its own11Even1 each
Math, 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.
Gemma 4 26B A4B 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 Gemma 4 26B A4B pulls ahead
- Follows unfamiliar, precisely checkable instructionsIFBench+30.7points ahead
- Olympiad math problems with short answersIMOAnswerBench+28.5points ahead
- Recent programming contest problemsLiveCodeBench v6+23.4points ahead
Where Kimi K2 Instruct pulls ahead
- Fixes real GitHub issues in Python projectsSWE-bench Verified+48.4points ahead
- Fixes real GitHub issues in many programming languagesSWE-bench Multilingual+30points ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+16.7points 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 Instruct
- SWE-bench VerifiedKimi K2 Instruct by 48.417.465.8+48.4
- SWE-bench MultilingualKimi K2 Instruct by 3017.347.3+30
- LiveCodeBench v6Gemma 4 26B A4B by 23.477.153.7+23.4
- SWE-bench ProKimi K2 Instruct by 13.913.827.7+13.9
- Terminal-Bench HardKimi K2 Instruct by 9.913.623.5+9.9
- SciCodeGemma 4 26B A4B by 9.34030.7+9.3
AgentsEven
- GDPValKimi K2 Instruct by 14.93.318.2+14.9
- BrowseCompGemma 4 26B A4B by 12.226.314.1+12.2
ReasoningGemma 4 26B A4B
- Humanity's Last ExamGemma 4 26B A4B by 12.919.36.4+12.9
- GPQA DiamondGemma 4 26B A4B by 2.579.276.7+2.5
- CritPttie00tie
FactsKimi K2 Instruct
- AA-Omniscience · Non-hallucinationKimi K2 Instruct by 16.713.630.3+16.7
- Vectara HHEM hallucination ratelower is betterGemma 4 26B A4B by 12.75.217.9+12.7
- AA-Omniscience · AccuracyKimi K2 Instruct by 6.319.125.4+6.3
Following instructionsGemma 4 26B A4B
- IFBenchGemma 4 26B A4B by 30.772.441.7+30.7
Other results18 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.
- HumanEvalKimi K2 Instruct by 43.35093.3+43.3
- τ²-Bench Telecom (AA run)Kimi K2 Instruct by 29.843.673.4+29.8
- AA Agentic IndexKimi K2 Instruct by 26.71137.7+26.7
- AA-OmniscienceKimi K2 Instruct by 24.2-50.8-26.6+24.2
- Artificial Analysis Coding IndexGemma 4 26B A4B by 13.439.325.9+13.4
- vectara_factual_consistencyGemma 4 26B A4B by 12.794.882.1+12.7
- MMLUKimi K2 Instruct by 11.777.889.5+11.7
- HLE (with tools)Kimi K2 Instruct by 9.717.226.9+9.7
- GPQA (unspecified)Gemma 4 26B A4B by 8.182.374.2+8.1
- vectara_avg_summary_lengthGemma 4 26B A4B by 7.967.159.2+7.9
- SuperGPQAGemma 4 26B A4B by 4.261.457.2+4.2
- τ²-BenchGemma 4 26B A4B by 2.468.265.8+2.4
- IFEvalGemma 4 26B A4B by 2.29289.8+2.2
- MATH-500 (EM)Kimi K2 Instruct by 1.795.797.4+1.7
- MMLU-ProGemma 4 26B A4B by 1.582.681.1+1.5
- AA IntelligenceGemma 4 26B A4B by 1.416.715.3+1.4
- vectara_answer_rateGemma 4 26B A4B by 1.299.898.6+1.2
- MMLU-Reduxtie92.792.7tie
Questions people ask
Which is better, Gemma 4 26B A4B or Kimi K2 Instruct?
Gemma 4 26B A4B wins four of the seven areas where both have results: reasoning, math, long documents and following instructions. Kimi K2 Instruct wins coding and facts. They are level on agents.
Which is better for coding?
Kimi K2 Instruct. It wins 4 of the 6 coding tests both models report; Gemma 4 26B A4B wins 2.
Which is cheaper?
Gemma 4 26B A4B costs $0.13 per million input tokens and $0.40 per million output tokens; Kimi K2 Instruct costs $0.60 and $2.50. That makes Gemma 4 26B A4B about 83% 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 17 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 18 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.