GLM-4.6 vs Kimi K2 Thinking

Benchmarks, API pricing and specs, head to head. Data updated 2026-06-10.

GLM-4.6

Z.ai (Zhipu) · Sep 2025

63.9

The budget coding-agent favorite of late 2025 — near-Sonnet coding utility with MIT weights and a $3/month plan.

Kimi K2 Thinking

Moonshot AI · Nov 2025

67.6

A trillion-parameter open reasoning agent that can chain 200–300 tool calls — the open-weights agentic standout of late 2025.

The verdict

Kimi K2 Thinking wins 6 of the 6 benchmarks these models share, against 0 for GLM-4.6. GLM-4.6 is about 1.1x cheaper per blended million tokens (3:1 input:output mix). Kimi K2 Thinking also takes 262K of context versus 200K for GLM-4.6.

Benchmark head-to-head 06

SWE-bench Verified
68%71.3%
GLM-4.6Kimi K2 Thinking
GPQA Diamond
~81%84.5%
GLM-4.6Kimi K2 Thinking
AIME 2025
~93.9%~94.5%
GLM-4.6Kimi K2 Thinking
HLE
~17.2%23.9%
GLM-4.6Kimi K2 Thinking
Terminal-Bench
40.5%~47.1%
GLM-4.6Kimi K2 Thinking
MMLU-Pro
~82.8%~84.6%
GLM-4.6Kimi K2 Thinking

Specs & pricing

GLM-4.6Kimi K2 Thinking
modhub Index63.967.6
Input price / 1M$0.6$0.6
Output price / 1M$2.2$2.5
Context window200K262K
Max output128K
Open weightsyes (MIT)yes (Modified MIT)
Reasoning modelyesyes
Multimodal inputtexttext
Knowledge cutoffJul 2025Apr 2025
ReleasedSep 2025Nov 2025
Example monthly cost*$9.30$9.75

* 10M input + 1.5M output tokens per month at list prices, no caching. Green = better value on that row.

Frequently asked questions

Which is better, GLM-4.6 or Kimi K2 Thinking?
Kimi K2 Thinking wins 6 of the 6 benchmarks these models share, against 0 for GLM-4.6. GLM-4.6 is about 1.1x cheaper per blended million tokens (3:1 input:output mix). Kimi K2 Thinking also takes 262K of context versus 200K for GLM-4.6.
Which is cheaper, GLM-4.6 or Kimi K2 Thinking?
GLM-4.6 costs $0.6/$2.2 per million input/output tokens, while Kimi K2 Thinking costs $0.6/$2.5. For a typical workload of 10M input and 1.5M output tokens per month, that's $9.30 versus $9.75.
Which model is better for coding, GLM-4.6 or Kimi K2 Thinking?
On SWE-bench Verified — the standard agentic-coding benchmark — Kimi K2 Thinking scores 71.3% versus 68% for GLM-4.6, making Kimi K2 Thinking the stronger pick for coding agents.

More comparisons