Noometry

Model comparison

Gemini 3.1 Pro Preview vs Kimi K2.5

Gemini 3.1 Pro Preview is the stronger model overall, scoring 56.7 to 47.8 on the Noometry Index. Kimi K2.5 costs 5.0× less per token, which makes it the better buy when Gemini 3.1 Pro Preview's lead doesn't matter for your workload.

Last verified . 46 shared benchmarks.

Gemini 3.1 Pro Preview Google

56.7

Rank #20 Confirmed

Kimi K2.5 Moonshot AI

47.8

Rank #50 Confirmed

Summary

  • They share 46 benchmarks with published results for both. Gemini 3.1 Pro Preview scores higher in 7 categories and Kimi K2.5 in 3 categories; 10 gaps are clear of the uncertainty.
  • The widest gap is in reasoning, where Gemini 3.1 Pro Preview leads 71.7 to 31.2.
  • The biggest single-benchmark swing is ARC-AGI-2: 77.1% for Gemini 3.1 Pro Preview and 11.8% for Kimi K2.5.
  • Kimi K2.5 is cheaper at $0.45 / $2.25 per million input/output tokens, against $2 / $12 for Gemini 3.1 Pro Preview.
  • Gemini 3.1 Pro Preview accepts more context: 1.05M tokens versus 262K.
  • Kimi K2.5 has downloadable open weights; the other is API-only.

Side by side

Gemini 3.1 Pro PreviewKimi K2.5
ProviderGoogleMoonshot AI
Noometry Index56.747.8
Released2026-02-192026-01-27
WeightsProprietaryOpen
Context window1.05M262K
Max output66K262K
Input $ / M tokens$2$0.45
Output $ / M tokens$12$2.25
Results tracked7151

Sponsored placements are available on pages like this one. Advertise on Noometry

Category by category

Coding Kimi K2.5 leads

Gemini 3.1 Pro Preview: 42.7 (#88), Kimi K2.5: 48.7 (#48)

BenchmarkGemini 3.1 Pro PreviewKimi K2.5
SWE-bench Verified75.6%73.8%
LMArena WebDev14461436
SciCode58.9%49%
WeirdML72.1%45.6%
LMArena Coding14831474
ALE-Bench1,161821.65
DeepSWE11.7%—
SWE-bench Verified (bash only)—70.8%
SWE-bench Multilingual—67.3%
GSO22.6%—
MirrorCode8.9%—
AlgoTune2.02—

Agentic & Tool Use Gemini 3.1 Pro Preview leads

Gemini 3.1 Pro Preview: 37.7 (#33), Kimi K2.5: 34.0 (#44)

BenchmarkGemini 3.1 Pro PreviewKimi K2.5
Terminal-Bench80.2%43.2%
Vending-Bench 23,7741,198
APEX-Agents35.3%—
τ²-bench Banking26%—
DeepResearch Bench47.8%—
OSWorld—63.3%
PostTrainBench22%—
BALROG57%—
ExploitBench26.1%—
GBAEval0.8%—
GDP.pdf17%—
LMArena Search1211—
METR Time Horizons77%—

Reasoning Gemini 3.1 Pro Preview leads

Gemini 3.1 Pro Preview: 71.7 (#9), Kimi K2.5: 31.2 (#74)

BenchmarkGemini 3.1 Pro PreviewKimi K2.5
ARC-AGI-277.1%11.8%
SimpleBench79.6%46.8%
NYT Connections (extended)97.4%69.9%
ARC-AGI-198%65.3%
CritPt17.7%3.1%
Chess Puzzles55%12%
EnigmaEval36.8%3.4%
Thematic Generalization79.4%69.4%
LMArena Hard Prompts14841454
Epoch Capabilities Index154.92148.05
Kagi LLM Benchmark—78.5%
EBR-Bench14.3%—
Mystery Game Puzzles34%—
DTBench97.1%—
LMCA53.8%—
ForecastBench59—

Math Gemini 3.1 Pro Preview leads

Gemini 3.1 Pro Preview: 61.8 (#28), Kimi K2.5: 50.3 (#46)

BenchmarkGemini 3.1 Pro PreviewKimi K2.5
MathArena Final-Answer Competitions86.5%62.3%
OTIS Mock AIME 2024-202595.6%92.2%
LMArena Math14851470
FrontierMath (Feb 2025 set)36.9%27.9%
FrontierMath Tier 4 (v1)16.7%4.2%
FrontierMath (Tiers 1-3)59.6%—
FrontierMath Tier 426.8%—
ProofBench26%—

Knowledge Gemini 3.1 Pro Preview leads

Gemini 3.1 Pro Preview: 71.8 (#3), Kimi K2.5: 53.6 (#52)

BenchmarkGemini 3.1 Pro PreviewKimi K2.5
GPQA Diamond94.4%87.6%
Humanity's Last Exam46.4%24.4%
SimpleQA Verified73.5%34.3%
Vectara Hallucination Rate10.4%14.2%
LMArena Expert14841466

Multimodal Kimi K2.5 leads

Gemini 3.1 Pro Preview: 38.1 (#60), Kimi K2.5: 40.4 (#37)

BenchmarkGemini 3.1 Pro PreviewKimi K2.5
LMArena Vision12961269
LMArena Document14441430
Blueprint-Bench 226.5%—
Furniture Assembly26.7%—

Multilingual Gemini 3.1 Pro Preview leads

Gemini 3.1 Pro Preview: 57.1 (#11), Kimi K2.5: 54.1 (#50)

BenchmarkGemini 3.1 Pro PreviewKimi K2.5
LMArena Non-English14751433
LMArena Chinese15281495
LMArena French14881454
LMArena German14911441
LMArena Japanese14941422
LMArena Korean14581412
LMArena Russian14961436
LMArena Spanish14801450

Instruction Following Gemini 3.1 Pro Preview leads

Gemini 3.1 Pro Preview: 77.1 (#31), Kimi K2.5: 75.5 (#60)

BenchmarkGemini 3.1 Pro PreviewKimi K2.5
LMArena Instruction Following14651431

Long Context Kimi K2.5 leads

Gemini 3.1 Pro Preview: 47.3 (#17), Kimi K2.5: 52.1 (#7)

BenchmarkGemini 3.1 Pro PreviewKimi K2.5
CL-bench20.8%19.3%
CL-bench Life16.9%13.2%
LMArena Longer Query14821445
Fiction.LiveBench—86.1%

Writing & Preference Gemini 3.1 Pro Preview leads

Gemini 3.1 Pro Preview: 66.3 (#36), Kimi K2.5: 65.3 (#51)

BenchmarkGemini 3.1 Pro PreviewKimi K2.5
LMArena Text14801445
LMArena Creative Writing14821423
EQ-Bench Creative Writing14911579
LMArena Multi-Turn14861444
EQ-Bench 41142—

Frequently asked questions

Is Gemini 3.1 Pro Preview better than Kimi K2.5?

Gemini 3.1 Pro Preview is the stronger model overall, scoring 56.7 to 47.8 on the Noometry Index. Kimi K2.5 costs 5.0× less per token, which makes it the better buy when Gemini 3.1 Pro Preview's lead doesn't matter for your workload.

Which is cheaper, Gemini 3.1 Pro Preview or Kimi K2.5?

Kimi K2.5 is cheaper. It lists at $0.45 per million input tokens and $2.25 per million output tokens; Gemini 3.1 Pro Preview lists at $2 and $12.

Is Gemini 3.1 Pro Preview or Kimi K2.5 better for coding?

Kimi K2.5 scores higher on coding benchmarks: 48.7 versus 42.7 in the Noometry coding category.

Which has the bigger context window?

Gemini 3.1 Pro Preview does, with 1.05M tokens against 262K.

How many benchmarks do Gemini 3.1 Pro Preview and Kimi K2.5 share?

46 benchmarks have published results for both models. Gemini 3.1 Pro Preview has 71 scored results on Noometry and Kimi K2.5 has 51.

Related comparisons

Go deeper