kimi-k2.7-code vs glm-5.2

Pricing, Performance & Features Comparison

kimi-k2.7-code

Authormoonshot

Context Length262K

Reasoning

Providers1

ReleasedJun 2026

Knowledge Cutoff–

License–

Moonshot’s coding- and agent-specialized model built on Kimi K2.6; thinking is always on.

Input$0.95

Output$4

Latency (p50)5.4s

Output Limit–

Function Calling

JSON Mode

InputText

OutputText

moonshot

in$0.95out$4cache$0.19–

Latency (24h)

Success Rate (24h)

glm-5.2

Authorzai

Context Length1M

Reasoning

Providers1

ReleasedJun 2026

Knowledge Cutoff–

License–

Z.ai’s flagship (744B MoE, MIT open weights) with a 1M-token context window and adjustable reasoning effort, built for long-horizon coding and agentic tasks.

Input$1.4

Output$4.4

Latency (p50)7.1s

Output Limit131K

Function Calling

JSON Mode

InputText

OutputText

zai

in$1.4out$4.4cache$0.26–