GLM-5

by Zhipu · chat

Zhipu's latest flagship 744B MoE model (40B active) with coding capabilities aligned to Claude Opus 4.5. Excels at long-horizon agentic planning and execution, 200K context. MIT license.

Pricing

Input: $0.95/M tokens · Output: $3.04/M tokens

Capabilities

Function Calling, JSON Mode, Streaming, Reasoning

Context: 200K tokens

Max output: 128K tokens

Routes: 2/2 healthy

Performance

TTFT: 13762ms · Latency: 22295ms · Throughput: 30.5 tok/s