AI_APICallError: The messages parameter is illegal. Please check the documentation. - Zhipu AI GLM-4.5

Zhipu AI GLM-4.5 regularly hangs mid edit and waits for a response after the message "AI_APICallError: The messages parameter is illegal. Please check the documentation."

Edit: I don't think it is mid-edit. It seems to happen after session compaction. And it seems to then start its memory again from the last session.

Only settings that are none default are z.ai's recommended setup for 'Coding Plan' on opencode. In ~/.config/opencode/opencode.json I have:

{
    "$schema": "https://opencode.ai/config.json",
    "provider": {
        "zhipuai": {
            "api": "https://api.z.ai/api/coding/paas/v4"
        }
    }
}

This issue might be a duplicate of existing issues. Please check:

Opencode with GLM 4.5 String issue #1880: GLM 4.5 String issue - Similar GLM 4.5 model problem with different error message but same hanging behavior
Add support for Z.ai GLM Coding Plan (Anthropic-compatible provider) #2431: Add support for Z.ai GLM Coding Plan - Discusses the same Z.ai Anthropic-compatible endpoint configuration you're using
Please adapt to the official GLM-4.5 direct connection version instead of Openrouter #1559: Please adapt to the official GLM-4.5 direct connection version - Related to GLM-4.5 connectivity issues and official API support

Feel free to ignore if none of these address your specific case.

if this was happening at random intervals (like after compacts) can you try latest? There were some fixes shipped today

I'm still getting the message in the latest version. It's a shame, because apart from that, glm-4.5 runs very fast and intelligently in opencode compared to other cli tools I'm testing. Sadly, this error makes it pretty unusable, as I can't leave it unattended, and need to re-prompt the AI constantly to get it to continue after this error. My workflow is to set all permissions to allow and run it sandboxed in a VM, then connect through a mosh session on my phone. I'm not the only person that uses this type of workflow.

I haven't tested but someone in the discord said:

use the anthropic endpoint and it will fix it

ex:

"zai": {
      "name": "zai-anth-base",
      "npm": "@ai-sdk/anthropic",
      "models": {
        "glm-4.5": {
          "id": "glm-4.5",
          "name": "GLM-4.5",
          "attachment": true,
          "tool_call": true,
          },
          "limit": {
            "context": 131072,
            "output": 98304
          },
        },
      "options": {
        "baseURL": "https://api.z.ai/api/anthropic/v1"
      },
    },

Yea, I just saw that and tested it. It solves the issue. Thanks! I won't close for now in case anyone has any tweaks or recommendations to offer. Otherwise, feel free to close.

But you now would lose the CoT and reasoning from the model.

OK so the Anthropic API doesn't suport reasoning? I think I'd rather put up with re-prompting it than lose reasoning for now.

Have been facing the same issue! And, I also don't want to lose reasoning at this point.

Same problem here, today I've been getting this error repeatedly in opencode v0.9.5

Where does it say you lose CoT reasoning? just curious maybe I missed it

Where does it say you lose CoT reasoning? just curious maybe I missed it

In the replies above from shekohex:

But you now would lose the CoT and reasoning from the model.

Yeah I meant like where does the provider say this

Yeah I meant like where does the provider say this

Nowhere in the docs, but looks like the OpenAI format shows the CoT/Thinking but using the Anthropic compatible API DOES NOT show it, or maybe it is missing it altogether.

would love to see this fixed as well, I am using GLM-4.5 heavily as well, with subscription. I didn't switch to anthropic, this model is better with reasoning.

Expected 'id' to be a string

I suspect that one is unrelated.

My best guess from diagnosing the problem is that it's an upstream issue with the inference software. The tool call template might not be able to handle hallucinated output from the model gracefully (i.e. when it makes up a tool name that wasn't provided in the request).

That guess is based on the fact that this occurs far more often when tools are restricted, or nearly guaranteed if I ask it to use a nonexistent tool.
Now try calling a made up tool: `foogah`.
User (03:35 PM)

The tool foogah doesn't exist. Available tools are: bash, edit, webfetch, glob, grep, list, read, write, todowrite, toread, task.
Build glm-4.5 (03:35 PM)

Try anyway. It should cause an error.
User (03:36 PM)

AI_InvalidResponseDataError: Expected 'id' to be a string.

oh shoot, you're right. I did reproduce it too. can we silently ignore this error or maybe report it to llm?

silently ignore this error

That depends on if there's something to salvage, but I think in this situation you see no streaming response at all.

Technically z.ai and other inference providers need to patch the bug and have the API behave more like OpenAI: GPT will happily return a hallucinated tool name in the response, leaving it up to the client (OpenCode) to deal with. Not sure what a good workaround here would be.

The chat template for GLM 4.5 is tricky, so I have some sympathy.

Just want to say thankyou for this fix. glm-4.5 coder plan with opencode and openai endpoint is now ridiculously powerful -- almost too good to be true. I'll enjoy it while it lasts and z.ai sustain these prices. opencode has the best cli implementation I've tested.

Though unfortunately "AI_InvalidResponseDataError: Expected 'id' to be a string." is still bringing it to a halt at times, but it's still an improvement.

It is worse than ever using 0.10.3

opencode v0.12.1
Zhipu AI GLM-4.5

Whatever I send (for example, a simple "hi") shows the error: AI_APICallError: API 调用参数有误，请检查文档。

@michabbb

@michabbb can u share your opencode.json?

@rekram1-node

@rekram1-node sure, but there´s nothing special in there, i only removed the mcps and one custom provider that is NOT used here....
here´s what's left:

{
  "$schema": "https://opencode.ai/config.json",
  "theme": "github",
  "model": "groq/moonshotai/kimi-k2-instruct-0905",
  "autoupdate": true,
  "permission": {
    "bash": {
      "rm": "ask"
    }
  },
  "agent": { "ask": { "mode": "primary", "tools": { "*": false }, "prompt": " " } },
}

@michabbb

@michabbb are you on their coding plan subscription?

@rekram1-node

@rekram1-node nope, just using an api key that is connected to a pre-pay account with some $ in it.
in hope i am not mixing things up, i am talking about using a key from: https://z.ai/manage-apikey/apikey-list

i am confused, because: https://models.dev/?sort=output-costper&order=asc&search=GLM-4.5

shows "z.ai" and "Zhipu AI" as two different providers but in opencode i can only choose "Zhipu AI" - but whats the difference here ? 🙈

Z.AI and Zhipu AI are marketed as two separate providers from the company Zhipu. Former facing international users, and the latter facing Chinese users. If you obtained the API key from https://z.ai, use the Z.AI provider.

@fwang

@fwang thanks a lot for clarification, i was wondering why i do see "Zhipu Ai" but not "Z.Ai", now i know. because i used the wrong one 🙈

@rekram1-node sorry for the noise! 😞

hey no problem

I am also seeing a lot of AI_APICallError: Invalid API parameter, please check the documentation. errors suddenly. not sure the cause

Navigation Menu

Repository navigation

AI_APICallError: The messages parameter is illegal. Please check the documentation. - Zhipu AI GLM-4.5 #2557

18 remaining items

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

AI_APICallError: The messages parameter is illegal. Please check the documentation. - Zhipu AI GLM-4.5 #2557

Description

Activity

github-actions commented on Sep 12, 2025

rekram1-node commented on Sep 13, 2025

jgwhite0111 commented on Sep 14, 2025

rekram1-node commented on Sep 14, 2025

jgwhite0111 commented on Sep 14, 2025

shekohex commented on Sep 14, 2025

jgwhite0111 commented on Sep 15, 2025

vinaychilukuri commented on Sep 15, 2025

0serna commented on Sep 16, 2025

rekram1-node commented on Sep 16, 2025

jgwhite0111 commented on Sep 16, 2025

rekram1-node commented on Sep 16, 2025

shekohex commented on Sep 16, 2025

aemr3 commented on Sep 16, 2025

18 remaining items

aemr3 commented on Sep 17, 2025

jdiaz5513 commented on Sep 17, 2025

jgwhite0111 commented on Sep 18, 2025

jgwhite0111 commented on Sep 18, 2025

brokechubb commented on Sep 20, 2025

michabbb commented on Sep 28, 2025

rekram1-node commented on Sep 28, 2025

michabbb commented on Sep 28, 2025

rekram1-node commented on Sep 28, 2025

michabbb commented on Sep 28, 2025

fwang commented on Sep 28, 2025

michabbb commented on Sep 28, 2025

rekram1-node commented on Sep 28, 2025

ShamanicArts commented on Oct 2, 2025

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions