Inference360
Moonshot AI

Kimi K2 Thinking

Kimi K2 Thinking is a model from Moonshot AI. 11 benchmarks count toward its score, in 5 categories.

availableShows if the model has enough results for an index.
IndexOverall score. 50 is the middle.48.5 ±6.9
CoverageShare of the index weight with results.80%
SpeedOutput tokens per second.34/s
Input / 1MUS dollars per 1M input tokens.$0.6
Output / 1MUS dollars per 1M output tokens.$2.5
ContextMaximum tokens in one request.262K
EloLMArena rating and rank.N/A

50 is the middle of the board. The range shows the doubt in the index.

CapabilitiesScore per category. 50 is the middle.

50 is the middle
AgenticMulti-step tasks with tools.
47.3
CodingCode writing and repair.
45.1
ReasoningLogic problems and puzzles.
49.4
MultimodalTasks with images and text.
N/A
KnowledgeFacts and expert knowledge.
50.8
MultilingualTasks in many languages.
N/A
InstructionTasks with strict rules in the prompt.
N/A
MathMath problems.
51.4

Results

11 counted
BenchmarkThe test name.CategoryThe capability that the test measures.ResultThe score from the publisher.IndexThis result on the index scale.RunThe settings of the run.DateDate of the result.Published byThe source of the result.
MGSMMultilingual90.1%——9 Jan 2026Vals AI
AIMEMath85.4%53.5—16 Apr 2026Vals AI
GPQA diamondKnowledge84.2%56.0——Epoch AI
OTIS Mock AIME 2024-2025Math83.1%58.0——Epoch AI
MMLU ProKnowledge81.1%48.2—1 Sept 2026Vals AI
GPQA DiamondKnowledge78.5%50.7—1 Sept 2026Vals AI
SWE-bench VerifiedCoding63.4%48.4mini-SWE-agent26 Feb 2026SWE-bench team
SWE-bench VerifiedCoding63.4%48.4mini-SWE-agent1 Sept 2026SWE-bench team
LiveCodeBenchCoding63.1%41.7—1 Sept 2026Vals AI
SWE-benchCoding60.2%45.8—1 Sept 2026Vals AI
Terminal-Bench 1.0Agentic40.0%45.4—12 Jan 2026Vals AI
Terminal-Bench 2.0Agentic37.1%49.1—4 Jun 2026Vals AI
FrontierMath-2025-02-28-PrivateMath21.4%50.6——Epoch AI
Chess PuzzlesReasoning20.0%49.4——Epoch AI
FrontierMath-Tier-4-2025-07-01-PrivateMath0.0%43.4——Epoch AI

11 benchmarks count, from 14 of 15 results. A grey row does not count. Too few models took that benchmark.

Sources

BenchLM benchmark aggregationUsed with attribution; per-benchmark results credited to their original authorsOpenRouter, collected directlyNo licence statedVals AI, collected directlyNo licence stated. Read from the public leaderboard and credited to Vals AIEpoch AI, collected directlyCC BY — free to use and redistribute with attributionSWE-bench team, collected directlyNo licence stated for the leaderboard. The harness repo is MIT

Same level, lower price

Qwen3.8-Flash-Next66.4 · Freedots3-note Preview65.5 · FreeApodex 1.1 Mini59.1 · Free

More from Moonshot AI

Kimi K372.0Kimi K2.660.9Kimi K2.7 Code59.5Kimi K2.5 (Reasoning)54.6Kimi K2.553.3