All insights

Model APIs

GLM-5.1 Release Date: Z.ai’s Official Date and API Notes

Z.ai’s GLM-5.1 announcement and release notes are dated April 7, 2026. See what that date confirms—and what API users should verify before production use.

Illustration for Glm-5.1 release date

Z.ai’s official GLM-5.1 announcement and release-note entry are dated April 7, 2026. That is the best-supported date to give, but the records cited here do not establish whether public access began earlier. [1][2]

Release date versus open-weight publication

Z.ai’s announcement is titled “GLM-5.1: Towards Long-Horizon Tasks” and dated April 7, 2026; its release notes list GLM-5.1 on the same date. These records establish the official announcement and release-note date, not necessarily the earliest date the model was publicly available. [1][2]

CometAPI separately reports that Z.ai open-sourced the full weights on April 7. Treat that as third-party reporting: the dated repository activity alone does not establish when the weights were first published. A Hacker News commenter recalled that the model had been public for about two weeks when a blog post was new, but that relative recollection does not establish an earlier calendar date. [3][4]

What API users should verify before production

Z.ai’s GLM-5.1 documentation lists a 200K context length, up to 128K maximum output, streaming, function calling, and structured output. These are documented model capabilities; confirm that they are supported for your intended API product and account before depending on them in production. [5]

Z.ai’s standard API documentation gives https://api.z.ai/api/paas/v4 as the general endpoint and specifies Bearer authentication. Its chat-completions documentation currently illustrates a request using glm-5.3, while Z.ai’s Coding Plan configuration uses the model code glm-5.1 and has a separate endpoint. Confirm the correct model identifier, endpoint, and API key for the product you intend to use rather than assuming Coding Plan settings apply to the standard API. [6][7][8]

For model selection, the available community reports are mixed and task-dependent, not controlled evidence of a general ranking. Test your own workload through the intended provider, including tool calls, context needs, latency, quotas, and failure handling.

Sources

GLM-5.1: Towards Long-Horizon Tasks — Z.ai

New Released — Z.AI Developer Documentation

How to Use GLM-5.1 API — CometAPI

GLM-5.1: Towards Long-Horizon Tasks — Hacker News discussion

GLM-5.1 Overview — Z.AI Developer Documentation

Introduction — Z.AI Developer Documentation

Chat Completion — Z.AI Developer Documentation

How to Switch Models — Z.AI Developer Documentation

Contact us