Compare AI models

Specifications
Parameters
1T>1T
Context window
256k256k
API pricing
Cheapest input
$0.49AtlasCloud—
Cheapest output
$2.50AtlasCloud—
$223AtlasCloud · int4—
Benchmarks
SWE-Bench Verified
76.8%75.3%
LiveCodeBench
85%85.9%
Humanity's Last Exam · with tools
50.2%49.8%
GPQA Diamond
87.6%87.4%
MMLU-Pro
87.1%85.7%
Overview
CompanyMoonshot AIQwen
Release dateJan 27 2026Jan 26 2026
AccessOpen WeightClosed
Model detailsView modelView model

Frequently asked questions

Kimi K2.5 leads Qwen3-Max-Thinking on 4 of the 5 benchmarks they both report. Qwen3-Max-Thinking shipped 1 day before Kimi K2.5, so benchmark comparisons should account for the intervening progress.

Kimi K2.5 has 1T parameters, while Qwen3-Max-Thinking has >1T. Context windows are 256k (Kimi K2.5) vs 256k (Qwen3-Max-Thinking). Kimi K2.5 is open weight, while Qwen3-Max-Thinking is closed.

On SWE-Bench Verified, Kimi K2.5 leads at 76.8% vs Qwen3-Max-Thinking at 75.3%. On LiveCodeBench, Qwen3-Max-Thinking leads at 85.9% vs Kimi K2.5 at 85%. On Humanity's Last Exam · with tools, Kimi K2.5 leads at 50.2% vs Qwen3-Max-Thinking at 49.8%. On GPQA Diamond, Kimi K2.5 leads at 87.6% vs Qwen3-Max-Thinking at 87.4%. On MMLU-Pro, Kimi K2.5 leads at 87.1% vs Qwen3-Max-Thinking at 85.7%.