GLM-5.1 (Z AI)

GLM-5.1 is a text-to-text large language model in the GLM family released by Z.AI, the company also known as Zhipu AI, as the direct successor to GLM-5. Its development lineage is explicitly confirmed by third-party tracking sources that mark GLM-5 as superseded in favor of the newer GLM-5.1 release, signaling a planned generational replacement rather than a parallel variant. The model is positioned as a general-purpose assistant with open weights available to the broader developer community.

The model's practical identity centers on strong coding performance paired with broad reasoning capability. In Z.AI's own coding evaluation that used Claude Code as the testing framework, GLM-5.1 scored 45.3 points, reaching roughly 94.6% of Claude Opus 4.6's 47.9-point result and representing a 28% jump over GLM-5's 35.4-point baseline, a sizeable single-generation gain in this benchmark. Z.AI highlighted GLM-5.1 as a high-cost-effective alternative for coding workflows, rolling it out first to GLM Coding Plan subscribers on March 27, 2026 before broader availability, and positioning the model for developers who want near-frontier coding ability without premium-tier pricing.

LLM Gatewayzai/glm-5.1glm

Open provider page Provider docs

Quick Info

Provider: LLM Gateway
Model key: zai/glm-5.1
Release date: Apr 7, 2026
Last updated: Apr 7, 2026
Input modalities
Output modalities
Capabilities

Cost

Input token cost: $1.40
Output token cost: $4.40

Limits

Output tokens: 128,000 tokens
Context window: 200,000 tokens

Transparent token rates

Compare GLM-5.1 (Z AI) pricing

Rates are shown per one million tokens. Combined means one million input plus one million output tokens.

Browse this family

Latest news about GLM-5.1 (Z AI)

LLM Gateway

Coverage

An agmazon.com guide positions GLM-5.1 by Z.ai (formerly Zhipu AI) as a major open-source milestone, claiming it delivers 94.6% of Claude Opus 4.6's performance and scoring 45.3 on a 113-task coding evaluation. The article describes a "paradigm shift toward Agentic Engineering" designed for complex systems engineering Technically, the guide details GLM-5.1's 744-billion-parameter Mixture-of-Experts architecture activating only 40–44B parameters per inference, a 200K input / 128K output context window, and a 256-expert sparse structure routing the top-8 with 5.9% sparsity. It references "DeepSeek Sparse Attention" (DSA) for memory-ef