GLM-4.7-Flash vs Kimi K2 Instruct
Wins 1 of 6 areas
Following instructions
Wins 2 of 6 areas
Facts · Long documents
Kimi K2 Instruct wins more areas, narrowly.GLM-4.7-Flash is cheaper and better at following instructions.
Scores updated · 25 tests both models report · How we compare
Where each one wins
Tests won in each of the six 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.
- FactsGetting facts right instead of making them up12Kimi K2 Instruct2 of 3 tests
- Long documentsFinding answers in very long texts01Kimi K2 Instruct1 of 1 test
- Following instructionsDoing exactly what it is asked10GLM-4.7-Flash1 of 1 test
- CodingWriting and fixing software22Even2 each
- AgentsCarrying out multi-step tasks on its own11Even1 each
- ReasoningHard problems that need careful thinking11Even1 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.
GLM-4.7-Flash costs 85% 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 Instruct pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+24.2points ahead
- Graduate-level biology, physics and chemistry questionsGPQA Diamond+18.6points ahead
- Reasons across sets of long documentsAA-LCR+11.3points ahead
Where GLM-4.7-Flash pulls ahead
- Finds hard-to-locate facts by browsing the webBrowseComp+28.7points ahead
- Follows unfamiliar, precisely checkable instructionsIFBench+19.1points ahead
- Recent programming contest problemsLiveCodeBench v6+10.3points ahead
Every test, side by side
All 25 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingEven
- LiveCodeBench v6GLM-4.7-Flash by 10.36453.7+10.3
- SWE-bench VerifiedKimi K2 Instruct by 6.659.265.8+6.6
- SciCodeGLM-4.7-Flash by 333.730.7+3
- Terminal-Bench HardKimi K2 Instruct by 1.52223.5+1.5
AgentsEven
- BrowseCompGLM-4.7-Flash by 28.742.814.1+28.7
- GDPValKimi K2 Instruct by 1.316.918.2+1.3
ReasoningEven
- GPQA DiamondKimi K2 Instruct by 18.658.176.7+18.6
- Humanity's Last ExamGLM-4.7-Flash by 1.27.66.4+1.2
- CritPttie0.30tie
FactsKimi K2 Instruct
- AA-Omniscience · Non-hallucinationKimi K2 Instruct by 24.26.130.3+24.2
- AA-Omniscience · AccuracyKimi K2 Instruct by 9.216.125.4+9.2
- Vectara HHEM hallucination ratelower is betterGLM-4.7-Flash by 8.69.317.9+8.6
Following instructionsGLM-4.7-Flash
- IFBenchGLM-4.7-Flash by 19.160.841.7+19.1
Other results11 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.
- AIME 2025GLM-4.7-Flash by 42.191.649.5+42.1
- AA-OmniscienceKimi K2 Instruct by 36-62.6-26.6+36
- τ²-Bench Telecom (AA run)GLM-4.7-Flash by 25.498.873.4+25.4
- τ²-BenchGLM-4.7-Flash by 13.779.565.8+13.7
- vectara_avg_summary_lengthGLM-4.7-Flash by 12.671.859.2+12.6
- vectara_factual_consistencyGLM-4.7-Flash by 8.690.782.1+8.6
- AA Agentic IndexGLM-4.7-Flash by 8.34637.7+8.3
- vectara_answer_rateKimi K2 Instruct by 791.698.6+7
- GPQA (unspecified)GLM-4.7-Flash by 175.274.2+1
- AA Intelligencetie14.915.3tie
- Artificial Analysis Coding Indextie25.925.9tie
Questions people ask
Which is better, GLM-4.7-Flash or Kimi K2 Instruct?
Kimi K2 Instruct wins two of the six areas where both have results: facts and long documents. GLM-4.7-Flash wins following instructions, and costs 85% less. They are level on coding, agents and reasoning.
Which is better for coding?
Neither. They win 2 coding tests each of the 4 both models report.
Which is cheaper?
GLM-4.7-Flash costs $0.07 per million input tokens and $0.40 per million output tokens; Kimi K2 Instruct costs $0.60 and $2.50. That makes GLM-4.7-Flash about 85% cheaper for the same work.
How do you compare the two?
We use the 25 benchmark tests both models have published scores on. The verdict counts the 14 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 11 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.