Inference
EN
Moonshot AI

Kimi K2 Thinking

Kimi K2 Thinking is a model from Moonshot AI. 10 benchmarks count toward its index, in 4 categories.

  • Open weights
  • Input: text
  • Released 2025-11
RankedEnough results for an index.
IndexAll categories, out of 100.48.2±8.6Median 51.0
CoverageIndex weight that has results.65%
SpeedOutput tokens per second.48/sMedian 67/s
First tokenSeconds to the first output token.1.8 sMedian 1.3 s
Input / 1MUS dollars per 1M input tokens.$0.60Median $0.60
Output / 1MUS dollars per 1M output tokens.$2.50Median $2.55
ContextMaximum tokens in one request.262KMedian 262K
EloLMArena rating and rank.—Median 1422

Details

From OpenRouter
API IDThe model ID on OpenRouter.
moonshotai/kimi-k2-thinking
Max outputMaximum output tokens per request.
98K
Knowledge cutoffLast date of training data.
—
ToolsTool calls on OpenRouter.
Yes
JSON outputOutput follows a given JSON schema.
Yes
Cache read / 1MUS dollars per 1M cached tokens.
$0.15
WeightsLink to the published weights.
Hugging Face (opens in a new tab) ↗
RetiresDate OpenRouter removes the model.
—

Where to run

APIs that serve this model.
4 providers
ProviderTypeMaker, cloud, host or router.InputOutputContext
Amazon BedrockDeprecatedCloud$0.60$2.50262K
IO.NETHost$0.60$2.5033K
Novita AIHost$0.60$2.50262K
OpenRouterRouter$0.60$2.50262K

US dollars per 1M tokens, list price. From models.dev (opens in a new tab) (MIT).

Run locally

1026.41B parameters
16-bit
2156.5 GB
8-bit
1146.1 GB
4-bit
654.4 GB
Weights
Hugging Face (opens in a new tab) ↗

Memory need with a short context. The 4-bit copy needs more than 128 GB. Compare local models

Capabilities

Index per category, out of 100.
Out of 100
AgenticMulti-step tasks with tools.2 counted48.0147 / 218
CodingCode writing and repair.2 counted44.9163 / 209
ReasoningLogic problems and puzzles.None counted——
MultimodalTasks with images and text.None counted——
KnowledgeFacts and expert knowledge.2 counted49.2140 / 268
MultilingualTasks in many languages.None counted——
InstructionTasks with strict prompt rules.None counted——
MathMath problems.4 counted50.585 / 168

Results

10 counted
BenchmarkThe benchmark name.ResultThe score from the source.PlacePlace among results on this benchmark.
SWE-bench Verified · swebench.comCoding63.4%8 / 46
GPQA diamond · Epoch AIKnowledge84.2% ±2.175 / 217
AIMEMath85.4% ±1.336 / 95
SWE-bench Verified · SWE-bench/experimentsCoding63.4%25 / 65
OTIS Mock AIME 2024-2025Math83.1% ±4.880 / 200
FrontierMath-2025-02-28-PrivateMath21.4% ±2.428 / 69
Terminal-Bench 1.0Agentic40.0% ±5.522 / 47
GPQA Diamond · Vals AIKnowledge78.5% ±2.272 / 135
Terminal-Bench 2.0 · Vals AIAgentic37.1% ±5.138 / 66
MMLU ProKnowledge81.1% ±0.484 / 135
LiveCodeBenchCoding63.1% ±1.2102 / 142
FrontierMath-Tier-4-2025-07-01-PrivateMath0.0%45 / 55
SWE-benchCoding60.2% ±2.277 / 87
FrontierMath-Tier-4-2025-07-01-PublicMath0.0%3 / 36
CaseLaw v2Legal65.7% ±1.810 / 53
MedQAHealthcare92.6% ±0.229 / 95
FrontierMath-2025-02-28-PublicMath30.0% ±16.322 / 64
CorpFin v2Finance60.6% ±1.066 / 131
TaxEval v2Finance71.7% ±0.974 / 143
MGSMMultilingual90.1% ±1.841 / 74
LegalBenchLegal80.2% ±0.589 / 149
Finance Agent v1.1Finance36.6% ±2.642 / 51

Sources

BenchLMCC BY-NC 4.0OpenRouterTerms restrict reuseVals AINo licence statedEpoch AICC BYSWE-bench teamNo licence statedmodels.devMIT

History

5 releases
Index49.2 → 48.2
23 Oct 202612 Nov 2026
Output / 1M$2.50 → $2.50
23 Oct 202612 Nov 2026

Questions

How much does Kimi K2 Thinking cost?
$0.60 per 1M input tokens, $2.50 per 1M output tokens.
How good is Kimi K2 Thinking?
Its index is 48.2 out of 100. That is place 159 of 283 ranked models.
What is Kimi K2 Thinking best at?
Math: 50.5 out of 100, place 85 of 168.
How fast is Kimi K2 Thinking?
48 output tokens per second, 1.8 s to the first token.
What is the context window of Kimi K2 Thinking?
262K tokens.
Is Kimi K2 Thinking open source?
Yes. Its weights are published.
When was Kimi K2 Thinking released?
November 2025.

Badge

Kimi K2 Thinking on Inference360: place 159 of 283, index 48.2
HTML
<a href="https://inference360.ai/models/moonshot-ai-kimi-k2-thinking"><img src="https://inference360.ai/badges/moonshot-ai-kimi-k2-thinking.svg" alt="Kimi K2 Thinking on Inference360: place 159 of 283, index 48.2" height="20"></a>
Markdown
[![Kimi K2 Thinking on Inference360: place 159 of 283, index 48.2](https://inference360.ai/badges/moonshot-ai-kimi-k2-thinking.svg)](https://inference360.ai/models/moonshot-ai-kimi-k2-thinking)

Compare with

Qwen3.5 Plus Thinking48.1GPT-5 mini48.4

Lower price

dots3-note Preview63.4 · FreeApodex 1.1 Mini55.7 · FreeNemotron 3 Ultra50.8 · Free

More from Moonshot AI

All models
Kimi K2 0905UnrankedKimi K2.5 (Reasoning)53.0Kimi K2.552.2Kimi K239.6Kimi K2.658.2