GLM-4.7Open Weight

Released

GLM-4.7 is an AI model released by Z.ai on Monday, Dec 22 2025, 83 days after GLM-4.6. It is an open-weight model — the trained weights are available to download and run. It has a 200k token context windowHow much information the model can work with in one request, measured in tokens — pieces of words or other data. This includes your question, attachments, conversation and reply. Limits can vary by model variant and provider.. Benchmark results (shown below) cover SWE-Bench Verified, LiveCodeBench, Terminal-Bench 2.0, BrowseComp, Humanity's Last Exam, and GPQA Diamond.

API pricing

Input
$0.60
Cached input
$0.11
Output
$2.20
  • Reasoning or thinking is supported.
USD per 1M tokensVerified October 19, 2026Read from docs.z.ai. Click the check to open the source.

Available from

DeepInfraflex$0.32$1.40200Kfp4
DeepInfra—$0.40$1.75200Kfp4
Venice—$0.4004$1.9292200Kfp4
Novita—$0.60$2.20200Kfp8
Z.AI—$0.60$2.20200Kfp4
Mancer 2—$0.70$2.50128Kfp4
USD per 1M tokensChecked November 15, 2026Third-party rates from openrouter.ai, not the lab’s own list price.

Benchmarks

Coding

SWE-Bench VerifiedCoding — Real coding tasks pulled from open-source projects — the AI has to find and fix actual bugs. A human-checked version of the original SWE-Bench. Higher is better.
73.8%
#34 of 80
LiveCodeBenchCompetitive coding — Coding problems published so recently the AI can't have seen them in training — a contamination-free test of raw programming skill. Higher is better.via BenchLM
84.9%
#10 of 36

Terminal & CLI

Terminal-Bench 2.0Agentic terminal coding — Can the AI work in a command-line terminal — running commands and finishing technical setup tasks the way a developer would? (Version 2.0 of the test.) Higher is better.
41%
#24 of 28

Agentic & tool use

BrowseCompWeb browsing — Can the AI browse the web and track down hard-to-find answers? Higher is better.
52%
#33 of 35

Reasoning & science

Humanity's Last ExamMultidisciplinary reasoning — Humanity's Last Exam — extremely hard expert questions across many subjects, written so you can't just look up the answer. “No tools” means the AI answers on its own. Higher is better.
24.8%no tools
42.8%with tools
no tools
#35 of 49
with tools
#37 of 60
GPQA DiamondScience — Graduate-level science questions in biology, physics, and chemistry — hard enough that subject-matter PhDs score around 65%. Higher is better.
85.7%
#41 of 101

Source: BenchLM, retrieved 24 August 2026. Other sources are identified on the linked benchmark pages.

Compare GLM-4.7 with

GLM-4.7

About

GLM-4.7 was released in December 2025 with a focus on coding, tool use and more consistent behavior during multi-turn tasks. Z.ai highlighted improvements over GLM-4.6 in repository work, terminal tasks and generating interfaces. The downloadable model continued the GLM line's use of reasoning to support an agent that acted through external tools.

The release introduced preserved thinking, which retained reasoning blocks across turns in coding-agent conversations, and turn-level controls over thinking behavior. These features addressed continuity as an assistant moved between decisions and actions. Z.ai reported 73.8% on SWE-bench Verified in its model-card evaluation; that result belongs to the stated agent setup rather than a guarantee about every coding framework or task.

Frequently asked questions

GLM-4.7 was released by Z.ai on Monday, Dec 22 2025.

All Z.ai releases

13 tracked