Add ZAI Coding Plan GLM-5.2 with Anthropic provider configuration

[ririnto]

Summary

• Add new provider-model definition for GLM-5.2 in providers/zai-coding-plan/models/glm-5.2.toml.
• Include provider-specific fields for Z.AI Coding Plan using @ai-sdk/anthropic with endpoint https://api.z.ai/api/anthropic.
• Set reasoning to true, tool_call to true, text-only input modality, context window 1_000_000, and max output 131_072.
• Set reasoning_options to high and max.

Notes

• The change to @ai-sdk/anthropic was made after checking Z.AI docs at https://docs.z.ai/devpack/latest-model.md, where the GLM-5.2 entry supports reasoning_options: ["high", "max"].
• No provider-agnostic model metadata (models/ entry) was added because the change is scoped to Z.AI Coding Plan configuration.

References

• Z.AI latest-model reference: https://docs.z.ai/devpack/latest-model.md
• PR/issue convention in this repository uses provider/model TOML additions as standalone metadata updates (for example, issue Add glm-4.6 for zhipuai, zhipuai-coding-plan, zai and zai-coding-plan #250 and Add GLM 5.1 to the list of models for NanoGPT #1291).

[augushong]

+1

[MistyMoonR]

+1

[katsugtgz]

+1

[Mxrk]

Shouldn’t it be glm-5.2[1m] as the docs are stating:
"Note: To enable GLM 1M context, add the [1m] suffix to the model name (e.g., glm-5.2[1m])“

[ririnto]

Shouldn’t it be glm-5.2[1m] as the docs are stating: "Note: To enable GLM 1M context, add the [1m] suffix to the model name (e.g., glm-5.2[1m])“

That's not part of model name. Just claude code's 1m context definition.

[ririnto]

Shouldn’t it be glm-5.2[1m] as the docs are stating: "Note: To enable GLM 1M context, add the [1m] suffix to the model name (e.g., glm-5.2[1m])“

That's not part of model name. Just claude code's 1m context definition.

For example, when use claude sonnet 1m cotext, set model at ~/.claude/settings.json like {"env": {"ANTHROPIC_DEFAULT_SONNET_MODEL": "claude-sonnet-4-6[1m]"}}. But model name is still "claude-sonnet-4-6".

[MistyMoonR]

Although official requests for GLM-5.1 are automatically routed to GLM-5.2, this model is mainly added to enable reasoning level configuration and expand the context window to 1,000,000 tokens.