GPT-5.5 vs Kimi K2.5
Wins 8 of 8 areas
Coding · Agents · Reasoning · Facts · Images and charts · Math · Long documents · Following instructions
Wins 0 of 8 areas
—
GPT-5.5 is the stronger all-rounder.Kimi K2.5 is cheaper.
Scores updated · 50 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.
- AgentsCarrying out multi-step tasks on its own60GPT-5.56 of 6 tests
- CodingWriting and fixing software50GPT-5.55 of 5 tests
- ReasoningHard problems that need careful thinking40GPT-5.54 of 4 tests
- FactsGetting facts right instead of making them up31GPT-5.53 of 4 tests
- MathCompetition and research-level math20GPT-5.52 of 2 tests
- Images and chartsUnderstanding pictures, charts and video21GPT-5.52 of 3 tests
- Long documentsFinding answers in very long texts10GPT-5.51 of 1 test
- Following instructionsDoing exactly what it is asked10GPT-5.51 of 1 test
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.
Kimi K2.5 costs 90% 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 GPT-5.5 pulls ahead
- Abstract visual puzzles that people can solveARC-AGI-2+73.2points ahead
- Command-line tasks in a real terminalTerminal-Bench 2.1+38.6points ahead
- Short factual questions, answered correctlySimpleQA Verified+28.7points ahead
Where Kimi K2.5 pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+23.3points ahead
- Math problems shown in pictures and chartsMathVista+5.9points ahead
Every test, side by side
All 50 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingGPT-5.5
- Terminal-Bench 2.1GPT-5.5 by 38.684.345.7+38.6
- Terminal-Bench HardGPT-5.5 by 25.860.634.8+25.8
- SWE-bench ProGPT-5.5 by 7.958.650.7+7.9
- LMArena · WebDevGPT-5.5 by 77 rating points15131436+77 rating
- SciCodeGPT-5.5 by 6.855.849+6.8
AgentsGPT-5.5
- AA ApexAgentsGPT-5.5 by 26.237.711.5+26.2
- GDPValGPT-5.5 by 25.542.717.2+25.5
- τ-Bench V3 · BankingGPT-5.5 by 24.83914.2+24.8
- OSWorld-VerifiedGPT-5.5 by 15.478.763.3+15.4
- MCP AtlasGPT-5.5 by 11.575.363.8+11.5
- BrowseCompGPT-5.5 by 9.584.474.9+9.5
ReasoningGPT-5.5
- ARC-AGI-2GPT-5.5 by 73.28511.8+73.2
- CritPtGPT-5.5 by 2427.13.1+24
- Humanity's Last ExamGPT-5.5 by 15.145.830.7+15.1
- GPQA DiamondGPT-5.5 by 5.693.587.9+5.6
FactsGPT-5.5
- SimpleQA VerifiedGPT-5.5 by 28.76334.3+28.7
- AA-Omniscience · Non-hallucinationKimi K2.5 by 23.31134.3+23.3
- AA-Omniscience · AccuracyGPT-5.5 by 22.85835.2+22.8
- Vectara HHEM hallucination ratelower is betterGPT-5.5 by 4.99.314.2+4.9
Images and chartsGPT-5.5
- MathVistaKimi K2.5 by 5.984.290.1+5.9
- MMMU-ProGPT-5.5 by 4.579.975.4+4.5
- LMArena · VisionGPT-5.5 by 23 rating points12921269+23 rating
MathGPT-5.5
- HMMT Feb. 2026GPT-5.5 by 11.498.587.1+11.4
- AIME 2026GPT-5.5 by 4.210095.8+4.2
Other results24 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.
- GDPval-AA v2GPT-5.5 by 500 rating points15091009+500 rating
- CyberGymGPT-5.5 by 40.581.841.3+40.5
- Terminal-Bench 2.0GPT-5.5 by 31.982.750.8+31.9
- frontiermath_tier_4_v1GPT-5.5 by 31.235.44.2+31.2
- ARC-AGI-1GPT-5.5 by 29.79565.3+29.7
- Artificial Analysis Coding IndexGPT-5.5 by 28.174.946.8+28.1
- AA-OmniscienceGPT-5.5 by 27.820.5-7.3+27.8
- Tool-DecathlonGPT-5.5 by 27.855.627.8+27.8
- ToolathlonGPT-5.5 by 27.855.627.8+27.8
- BabyVisionGPT-5.5 by 19.455.936.5+19.4
- NL2RepoGPT-5.5 by 18.750.732+18.7
- AA Agentic IndexGPT-5.5 by 15.637.321.7+15.6
- AA IntelligenceGPT-5.5 by 14.938.423.5+14.9
- matharena_visual_math_overallGPT-5.5 by 14.394.980.6+14.3
- SimpleVQAKimi K2.5 by 12.658.671.2+12.6
- WorldVQAKimi K2.5 by 11.734.646.3+11.7
- LiveBenchGPT-5.5 by 11.680.769.1+11.6
- vectara_answer_rateGPT-5.5 by 7.810092.2+7.8
- HMMT Nov. 2025GPT-5.5 by 7.396.589.2+7.3
- vectara_factual_consistencyGPT-5.5 by 4.990.785.8+4.9
- ResearchRubricsGPT-5.5 by 4.56459.5+4.5
- HLE (with tools)GPT-5.5 by 252.250.2+2
- τ²-Bench Telecom (AA run)Kimi K2.5 by 293.995.9+2
- vectara_avg_summary_lengthGPT-5.5 by 17.6 rating points129.6112+17.6 rating
Questions people ask
Which is better, GPT-5.5 or Kimi K2.5?
GPT-5.5 wins all eight areas we test: coding, agents, reasoning, facts, images and charts, math, long documents and following instructions. Kimi K2.5 wins none, but costs 90% less.
Which is better for coding?
GPT-5.5. It wins 5 of the 5 coding tests both models report; Kimi K2.5 wins none.
Which is cheaper?
GPT-5.5 costs $5.00 per million input tokens and $30.00 per million output tokens; Kimi K2.5 costs $0.60 and $3.00. That makes Kimi K2.5 about 90% cheaper for the same work.
How do you compare the two?
We use the 50 benchmark tests both models have published scores on. The verdict counts the 26 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 24 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.