GPT-5.2 Codex vs Kimi K2.6
Wins 4 of 8 areas
Agents · Math · Long documents · Following instructions
Wins 3 of 8 areas
Coding · Reasoning · Images and charts
GPT-5.2 Codex wins more areas, narrowly.Kimi K2.6 is cheaper and better at reasoning.
Scores updated · 28 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.
- Following instructionsDoing exactly what it is asked20GPT-5.2 Codex2 of 2 tests
- AgentsCarrying out multi-step tasks on its own10GPT-5.2 Codex1 of 1 test
- MathCompetition and research-level math10GPT-5.2 Codex1 of 1 test
- Long documentsFinding answers in very long texts10GPT-5.2 Codex1 of 1 test
- ReasoningHard problems that need careful thinking03Kimi K2.63 of 4 tests · 1 tie
- CodingWriting and fixing software34Kimi K2.64 of 7 tests
- Images and chartsUnderstanding pictures, charts and video01Kimi K2.61 of 1 test
- FactsGetting facts right instead of making them up11Even1 each
Agents, images and charts, 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 K2.6 costs 69% less for the same work.
The biggest differences
The three tests each model wins by the widest margin. Scores are out of 100.
Where GPT-5.2 Codex pulls ahead
- Real work tasks from 44 professionsGDPVal+12.4points ahead
- Answers hard knowledge questions correctlyAA-Omniscience · Accuracy+8.5points ahead
- Programming problems, refreshed regularlyLiveBench · Coding+5points ahead
Where Kimi K2.6 pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+32.9points ahead
- Fixes real GitHub issues in Python projectsSWE-bench Verified+7.4points ahead
- Hard command-line tasks in a real terminalTerminal-Bench Hard+6.8points ahead
Every test, side by side
All 28 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingKimi K2.6
- LMArena · WebDevKimi K2.6 by 170 rating points13391509+170 rating
- SWE-bench VerifiedKimi K2.6 by 7.472.880.2+7.4
- Terminal-Bench HardKimi K2.6 by 6.837.143.9+6.8
- LiveBench · CodingGPT-5.2 Codex by 583.678.6+5
- SciCodeGPT-5.2 Codex by 3.154.651.5+3.1
- LiveBench · Agentic CodingGPT-5.2 Codex by 2.549.446.9+2.5
- SWE-bench ProKimi K2.6 by 2.256.458.6+2.2
ReasoningKimi K2.6
- Humanity's Last ExamKimi K2.6 by 1.835.737.5+1.8
- LiveBench · ReasoningKimi K2.6 by 1.777.779.4+1.7
- GPQA DiamondKimi K2.6 by 1.289.991.1+1.2
- CritPttie8.78tie
FactsEven
- AA-Omniscience · Non-hallucinationKimi K2.6 by 32.926.659.5+32.9
- AA-Omniscience · AccuracyGPT-5.2 Codex by 8.541.132.6+8.5
Following instructionsGPT-5.2 Codex
- LiveBench · Instruction FollowingGPT-5.2 Codex by 2.166.564.4+2.1
- IFBenchGPT-5.2 Codex by 1.677.676+1.6
Other results9 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 Agentic IndexGPT-5.2 Codex by 34.456.522.1+34.4
- Artificial Analysis Coding IndexKimi K2.6 by 18.84361.8+18.8
- livebench_data_analysisGPT-5.2 Codex by 13.178.265.1+13.1
- AA-OmniscienceKimi K2.6 by 7.5-2.25.3+7.5
- τ²-Bench Telecom (AA run)Kimi K2.6 by 3.892.195.9+3.8
- Terminal-Bench 2.0Kimi K2.6 by 2.76466.7+2.7
- LiveBenchGPT-5.2 Codex by 2.174.372.2+2.1
- AA IntelligenceGPT-5.2 Codex by 1.528.527+1.5
- livebench_languageKimi K2.6 by 1.473.775.1+1.4
Questions people ask
Which is better, GPT-5.2 Codex or Kimi K2.6?
GPT-5.2 Codex wins four of the eight areas we test: agents, math, long documents and following instructions. Kimi K2.6 wins coding, reasoning and images and charts, and costs 69% less. They are level on facts.
Which is better for coding?
Kimi K2.6. It wins 4 of the 7 coding tests both models report; GPT-5.2 Codex wins 3.
Which is cheaper?
GPT-5.2 Codex costs $1.75 per million input tokens and $14.00 per million output tokens; Kimi K2.6 costs $0.95 and $4.00. That makes Kimi K2.6 about 69% cheaper for the same work.
How do you compare the two?
We use the 28 benchmark tests both models have published scores on. The verdict counts the 19 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 9 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.