ZMIME
Moonshot AI · Released Nov 6, 2025

Kimi K2 Thinking

Thinking Kimi model for slower research passes, planning, and hard technical questions.

  • Open weights
  • Reasoning
  • Tool calling
Capability
145.5ECI · #90 of 158
Input
$0.60per 1M tokens
Output
$2.50per 1M tokens
Context
262K262K max output
Pricing

What it costs

Input
$0.60per million tokens
Output
$2.50per million tokens
Blended (3:1)
$1.07pricier than 60% of priced models

Typical monthly bills

Monthly usageEstimated cost
Side project2M input + 0.5M output tokens$2.45
Team assistant25M input + 5M output tokens$27.50
Production app250M input + 50M output tokens$275.00

Median of 14 paid API providers on models.dev.

Capability

Independent benchmarks

145.5Epoch Capabilities Index
#90 of 158 scored models
68Median 147174

The shaded band is Epoch AI’s confidence range (143.7–147.3); the tick marks the median scored model.

  • GPQA DiamondGraduate-level science questions84.2%
  • OTIS Mock AIME 2024–2025Competition mathematics83.1%

Scores from Epoch AI, run independently of Moonshot AI.

Specifications

The details

Lab
Moonshot AI
Released
Nov 6, 2025
Knowledge cutoff
Aug 2024
Context window
262,144 tokens
Max output
262,144 tokens
Inputs
Text
Output
Text
Reasoning
Always on
Tool calling
Yes
Structured output
No
Availability
15 API providerslisted on models.dev
Lab-reported

What Moonshot AI claims

Published by the lab at launch. Settings vary, so compare these only with care.

BenchmarkScoreSettingSource
SWE-Bench Verified71.3 resolved—Source
Same lab

More from Moonshot AI

Questions

About Kimi K2 Thinking

How much does Kimi K2 Thinking cost?

Kimi K2 Thinking costs $0.60 input / $2.50 output per million tokens (median across 14 API providers). At a 3:1 input-to-output mix that is $1.07 per million tokens, more expensive than 60% of the 469 priced models we track.

What is the context window of Kimi K2 Thinking?

Kimi K2 Thinking accepts up to 262,144 tokens per request and can write up to 262,144 tokens in one response.

How good is Kimi K2 Thinking?

Epoch AI gives Kimi K2 Thinking a Capabilities Index score of 145.5 (likely range 143.7–147.3), ranking it #90 of 158 models Epoch has scored. Epoch AI benchmark results: GPQA Diamond 84.2%, OTIS Mock AIME 2024–2025 83.1%.

Is Kimi K2 Thinking open source?

Yes. Moonshot AI publishes the weights, so it can be downloaded and self-hosted.

What inputs does Kimi K2 Thinking support?

Kimi K2 Thinking accepts text and replies in text. It is a reasoning model, supports tool calling.

When was Kimi K2 Thinking released?

Moonshot AI released Kimi K2 Thinking on Nov 6, 2025. Its training data runs to Aug 2024.

What are the best alternatives to Kimi K2 Thinking?

The closest current models from other labs on capability, price and release date are GLM-4.7 (Z.ai (Zhipu), ECI 143.4, $0.60 / $2.20), GPT-5 Mini (OpenAI, ECI 146.0, $0.25 / $2.00), Qwen3.5 Plus (Alibaba (Qwen), ECI 147.3, $0.40 / $2.40) and Nemotron 3 Ultra 550B A55B (NVIDIA, ECI 147.0, $0.55 / $2.40).