Kimi K2 Thinking

Training compute

4.2×10²⁴ FLOP

Parameters

1T

Published

Nov 6, 2025

Kimi K2 Thinking is an AI model developed by Moonshot (China), first published in November 2025. It works in the language domain, on tasks such as language modeling/generation, question answering, search and 2 more.

Training it took an estimated 4.2×10²⁴ FLOP of compute (estimation method: comparison with other models). The model has 1,000,000,000,000 parameters.

Access: Open weights (restricted use). Its weights are openly available. It is built on top of Kimi K2. Epoch AI rates the confidence of this record as likely.

Full record

Organization: Moonshot
Country of organization: China
Domain: Language
Task: Language modeling/generation, Question answering, Search, Quantitative reasoning, Code generation
Training compute: 4.2×10²⁴ FLOP
Compute estimation method: Comparison with other models
Parameters: 1,000,000,000,000
Model accessibility: Open weights (restricted use)
Open weights: Yes
Base model: Kimi K2
Epoch confidence: Likely

Benchmark results

01	ALE Bench	Performance	597.5
02	Epoch Capabilities Index	ECI Score	146
03	METR Time Horizons	average_score	0.59
04	WeirdML	Accuracy	0.43
05	Terminal-Bench	Accuracy mean	0.36
06	CL Bench	Overall	0.12
07	Posttrainbench	Average (%)	0.07
08	Apex Agents	Pass@1 score	0.04