GLM 4.7 Max sub is slow and dumb

Description

The free version was actually pretty good for tool calling and orchestration. Now that I'm using the z.ai max sub, it feels slow and dumb. That has to be a bug.

Is there some config that was applied to the free model by default, such as thinking levels (don't see any high thinking variants in opencode) or anything else? Using GLM 4.7 max sub feels like significant downgrade to the free version.

Plugins

No response

OpenCode version

No response

Steps to reproduce

No response

Screenshot and/or share link

No response

Operating System

No response

Terminal

No response

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

Differences between the free and paid versions of GLM 4.7 #6074: Differences between the free and paid versions of GLM 4.7

Feel free to ignore if this doesn't address your specific case.

I'm wondering if it's related to this:

From docs.z.ai

Preserved Thinking

This capability is enabled by default on the Coding Plan endpoint and disabled by default on the standard API endpoint. If you want to enable Preserved Thinking in your product (primarily recommended for coding/agent scenarios), you can turn it on for the API endpoint by setting “clear_thinking”: false, and you must return the complete, unmodified reasoning_content back to the API.
All consecutive reasoning_content blocks must exactly match the original sequence generated by the model during the initial request. Do not reorder or edit these blocks; otherwise, performance may degrade and cache hit rates may be affected.

Turn-level Thinking

“Turn-level Thinking” is a capability that lets you control reasoning computation on a per-turn basis: within the same session, each request can independently choose to enable or disable thinking. This is a new capability introduced in GLM-4.7, with the following advantages:

More flexible cost/latency control: For lightweight turns like “asking a fact” or “tweaking wording,” you can disable thinking to get faster responses; for heavier tasks like “complex planning,” “multi-constraint reasoning,” or “code debugging,” you can enable thinking to improve accuracy and stability.

Smoother multi-turn experience: The thinking switch can be toggled at any point within a session. The model stays coherent across turns and keeps a consistent output style, making it feel “smarter when things are hard, faster when things are simple.”

Better for agent/tool-use scenarios: On turns that require quick tool execution, you can reduce reasoning overhead; on turns that require making decisions based on tool results, you can turn on deeper thinking—dynamically balancing efficiency and quality.

Same here. Very slow and very dumb

I subscribed for the Max for one year after a worse than Claude performance but better pricing using GLM 4.6. Since 4.7 it's just obnoxious how bad this is performing. It's slow as f*** and not performing nearly as close as it used to be. I feel cheated.

On contacting Z.ai when I received some errors about rate-limiting they replied with the following:

However, we notice that you receive error codes in Chinese, which is un-expected. The api endpoint you are using 'https://bigmodel.cn/claude-code' is incorrect, since you are a user in z.ai platform, please ensure to use the correct api endpoint with domain name api.z.ai.
For Claude Code: https://api.z.ai/api/anthropic
For Other Tools: https://api.z.ai/api/coding/paas/v4

I seemed to be able to increase performance with manually putting it in the config:

    "zhipuai-coding-plan": {
      "options": {
        "baseURL": "https://api.z.ai/api/coding/paas/v4"
      },
      "models": {
        "glm-4.7": {
          "limit": {
            "context": 100000,
            "output": 4096
          }
        },
        "glm-4.5": {
          "limit": {
            "context": 100000,
            "output": 4096
          }
        },
        "glm-4.5-air": {
          "limit": {
            "context": 100000,
            "output": 4096
          }
        },
        "glm-4.5-flash": {
          "limit": {
            "context": 100000,
            "output": 4096
          }
        },
        "glm-4.5v": {
          "limit": {
            "context": 100000,
            "output": 4096
          }
        },
        "glm-4.6": {
          "limit": {
            "context": 100000,
            "output": 4096
          }
        },
        "glm-4.6v": {
          "limit": {
            "context": 100000,
            "output": 4096
          }
        },
        "glm-4.6v-flash": {
          "limit": {
            "context": 100000,
            "output": 4096
          }
        }
      }
    },
    "zai-coding-plan": {
      "options": {
        "baseURL": "https://api.z.ai/api/coding/paas/v4"
      },
      "models": {
        "glm-4.7": {
          "limit": {
            "context": 100000,
            "output": 4096
          }
        },
        "glm-4.7-flash": {
          "limit": {
            "context": 100000,
            "output": 4096
          }
        },
        "glm-4.6": {
          "limit": {
            "context": 100000,
            "output": 4096
          }
        },
        "glm-4.6v": {
          "limit": {
            "context": 100000,
            "output": 4096
          }
        },
        "glm-4.5": {
          "limit": {
            "context": 100000,
            "output": 4096
          }
        },
        "glm-4.5-air": {
          "limit": {
            "context": 100000,
            "output": 4096
          }
        },
        "glm-4.5-flash": {
          "limit": {
            "context": 100000,
            "output": 4096
          }
        },
        "glm-4.5v": {
          "limit": {
            "context": 100000,
            "output": 4096
          }
        }
      }
    }
  },

I reduced the context as well for faster speed.

Would it by default be connecting to Chinese servers?

I also don't know what's the difference between zai-coding-plan and zhipuai-coding-plan...

When removing my API key it really seems to be a Chinese endpoint as per error message: 您的GLM Coding Plan套餐已到期，暂无法使用，前往官方续订后即可恢复https://bigmodel.cn/claude-code。

I will do some more testing hopefully it's resolved.

Max plan user as well. The config worked for me too and opencode is zippy as well. I know on GLM docs they make mention to use the specific URL so I thought it was default on opencode. Thank you!

What also helped me was adding in concurrency settings for all the models so opencode can make use of its queues.

Zhipu is the China branch of zai.

Right, is this not accounted for in OpenCode fault auth flow for z?

Instructions on Z website says default opencode auth flow works: https://docs.z.ai/devpack/tool/opencode

@TotalLag

@TotalLag could you please share the settings you use? specifically the part about concurrency.

@TotalLag

as per config:

    "zai-coding-plan": {
      "options": {
        "baseURL": "https://api.z.ai/api/coding/paas/v4",
        "concurrency": 5
      },

@TotalLag what number did you define and are you getting good results? Mine still aren't that great compared to Claude Sonnet using GH Copilot...

I checked my outgoing network traffic, and my endpoint is not pointing to the Chinese servers, but to the correct international servers (I didn't add the z.ai models to my config).
So that's definitely not the issue for me.

In opencode if you are using zai coding plan we use:

https://api.z.ai/api/coding/paas/v4

If you are using Zhipu coding plan we use:

https://open.bigmodel.cn/api/coding/paas/v4

So if your config is like this:
"zai-coding-plan": {
"options": {
"baseURL": "https://api.z.ai/api/coding/paas/v4"
},

It should make 0 difference

does setting concurrency up help? is there thinking modes to configure manually? I have no special config for my z.ai subscription, just connected through OC provider auth and let it add whatever its adding for it to configure. should I use my own config?

@mikaelj

I seemed to be able to increase performance with manually putting it in the config:

"zhipuai-coding-plan": {
  "options": {
    "baseURL": "https://api.z.ai/api/coding/paas/v4"
  },
  "models": {
    "glm-4.7": {
      "limit": {
        "context": 100000,
        "output": 4096
      }
    },
    "glm-4.5": {
      "limit": {
        "context": 100000,
        "output": 4096
      }
    },
    "glm-4.5-air": {
      "limit": {
        "context": 100000,
        "output": 4096
      }
    },
    "glm-4.5-flash": {
      "limit": {
        "context": 100000,
        "output": 4096
      }
    },
    "glm-4.5v": {
      "limit": {
        "context": 100000,
        "output": 4096
      }
    },
    "glm-4.6": {
      "limit": {
        "context": 100000,
        "output": 4096
      }
    },
    "glm-4.6v": {
      "limit": {
        "context": 100000,
        "output": 4096
      }
    },
    "glm-4.6v-flash": {
      "limit": {
        "context": 100000,
        "output": 4096
      }
    }
  }
},
"zai-coding-plan": {
  "options": {
    "baseURL": "https://api.z.ai/api/coding/paas/v4"
  },
  "models": {
    "glm-4.7": {
      "limit": {
        "context": 100000,
        "output": 4096
      }
    },
    "glm-4.7-flash": {
      "limit": {
        "context": 100000,
        "output": 4096
      }
    },
    "glm-4.6": {
      "limit": {
        "context": 100000,
        "output": 4096
      }
    },
    "glm-4.6v": {
      "limit": {
        "context": 100000,
        "output": 4096
      }
    },
    "glm-4.5": {
      "limit": {
        "context": 100000,
        "output": 4096
      }
    },
    "glm-4.5-air": {
      "limit": {
        "context": 100000,
        "output": 4096
      }
    },
    "glm-4.5-flash": {
      "limit": {
        "context": 100000,
        "output": 4096
      }
    },
    "glm-4.5v": {
      "limit": {
        "context": 100000,
        "output": 4096
      }
    }
  }
}

},
I reduced the context as well for faster speed.

Would it by default be connecting to Chinese servers?

I also don't know what's the difference between zai-coding-plan and zhipuai-coding-plan...

When removing my API key it really seems to be a Chinese endpoint as per error message: 您的GLM Coding Plan套餐已到期，暂无法使用，前往官方续订后即可恢复https://bigmodel.cn/claude-code。

I will do some more testing hopefully it's resolved.

@mikaelj After I used your parameters, I feel the speed has increased a lot. But should 128k context be the standard?
By the way, what does the output parameter do? Is it used to control the number of output tokens?

@TotalLag could you please share the settings you use? specifically the part about concurrency.

I am using the numbers z.ai gives from the dashboard. However they noted that it's API and not Coding Plan so I'm unsure if it's actually effective or placebo, but I've seen in logs where my agents will wait so I assume the concurrency is working. I just noticed it's not core opencode so ymmv:

oh-my-opencode.json:

{
	"$schema": "https://raw.githubusercontent.com/code-yeongyu/oh-my-opencode/master/assets/oh-my-opencode.schema.json",
	"agents": {
		"sisyphus": {
			"model": "zai-coding-plan/glm-4.7"
		},
		"hephaestus": {
			"model": "zai-coding-plan/glm-4.7"
		},
		"librarian": {
			"model": "zai-coding-plan/glm-4.5-air"
		},
		"explore": {
			"model": "zai-coding-plan/glm-4.5-air"
		},
		"oracle": {
			"model": "zai-coding-plan/glm-4.7"
		},
		"frontend-ui-ux-engineer": {
			"model": "zai-coding-plan/glm-4.6v"
		},
		"ultrabrain": {
			"model": "openrouter/openai/5.2-codex"
		},
		"document-writer": {
			"model": "zai-coding-plan/glm-4.5-air"
		},
		"multimodal-looker": {
			"model": "zai-coding-plan/glm-4.6v"
		}
	},
	"categories": {
		"quick": {
			"model": "zai-coding-plan/glm-4.5-air"
		},
		"writing": {
			"model": "zai-coding-plan/glm-4.5-air"
		},
		"unspecified-low": {
			"model": "zai-coding-plan/glm-4.5-air"
		},
		"artistry": {
			"model": "zai-coding-plan/glm-4.6v"
		},
		"visual-engineering": {
			"model": "zai-coding-plan/glm-4.6v"
		}
	},
	"background_task": {
		"modelConcurrency": {
			"zai-coding-plan/glm-4.6": 1,
			"zai-coding-plan/glm-4.6v-flashx": 3,
			"zai-coding-plan/glm-4.7": 2,
			"zai-coding-plan/glm-image": 1,
			"zai-coding-plan/glm-4.5": 10,
			"zai-coding-plan/glm-4.6v": 10,
			"zai-coding-plan/glm-4.7-flash": 1,
			"zai-coding-plan/glm-4.7-flashx": 3,
			"zai-coding-plan/glm-4-plus": 20,
			"zai-coding-plan/glm-4.5v": 10,
			"zai-coding-plan/glm-4.6v-flash": 1,
			"zai-coding-plan/autoglm-phone-multilingual": 5,
			"zai-coding-plan/glm-4.5-air": 5,
			"zai-coding-plan/glm-4.5-airx": 5,
			"zai-coding-plan/glm-4.5-flash": 2,
			"zai-coding-plan/glm-4-32b-0414-128k": 15,
			"zai-coding-plan/cogview-4-250304": 5,
			"zai-coding-plan/glm-asr-2512": 5,
			"zai-coding-plan/viduq1-text": 5,
			"zai-coding-plan/viduq1-image": 5,
			"zai-coding-plan/viduq1-start-end": 5,
			"zai-coding-plan/vidu2-image": 5,
			"zai-coding-plan/vidu2-start-end": 5,
			"zai-coding-plan/vidu2-reference": 5,
			"zai-coding-plan/cogvideox-3": 1
		}
	}
}

Slow and eats credit like a kid in candy store. Especially with the new weekly and hourly quotas

I'm really getting annoyed about how painfully slow the z.ai coding plan is. It often just hangs and I have to interrupt and tell it to "continue".
I don't have much ope that this gets better. Do you guys have any other recommendations except the 200 bucks per month plans?

@KyrillGobber

@KyrillGobber , I saw another post on which the author stated he's happy with Ollama Cloud. I think it's worth to try the $20.00 subscription and move on with $100.00 if the service is stable. z.ai support is also unresponsive after getting the initial ticket creation process. I sent them an email regarding the rate limiting I started seeing even though the plan doesn't mention anything beyond the weekly quota but I never heard back from them.

I did some benchmarking and it looks like GLM 5 comes in at about 40tps
GLM 4.7 is about 90tps
The surprise is GLM 4.7 Flash which comes in low at 30tps, and often has delayed starts.

The GLM models have stabilized for me in the past week or two.

EDIT: note that I’m on Z.ai’s max plan

I am on the same plan. I finally got a generic response from their support team and while I was reading it, I got connection reset by peer.

Following the recent release of our new model, we have seen a significant surge in global user traffic.As our model operates as a shared service available to users worldwide, our servers have been running under continuous high load 24/7.

Under these circumstances, you may occasionally experience issues such as slower response or concurrency errors. The root cause is limited server resources under extremely high demand.

Unfortunately, there is no immediate solution that can be applied from the client side at this moment.Our team is working diligently to expand server capacity, optimize model algorithms to reduce server load, and deliver a more stable and reliable experience for all users.

We sincerely appreciate your understanding and patience.

To stay organized issues are automatically closed after 60 days of no activity. If the issue is still relevant please open a new one.

Navigation Menu

Repository navigation

GLM 4.7 Max sub is slow and dumb #10738

Description

Plugins

OpenCode version

Steps to reproduce

Screenshot and/or share link

Operating System

Terminal

Preserved Thinking

Turn-level Thinking

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

GLM 4.7 Max sub is slow and dumb #10738

Description

Description

Plugins

OpenCode version

Steps to reproduce

Screenshot and/or share link

Operating System

Terminal

Activity

github-actions commented on Jan 26, 2026

moonray commented on Jan 27, 2026

Preserved Thinking

Turn-level Thinking

vladlearns commented on Jan 27, 2026

maarteNNNN commented on Jan 29, 2026

maarteNNNN commented on Jan 29, 2026

TotalLag commented on Feb 1, 2026

diegohb commented on Feb 2, 2026

mikaelj commented on Feb 2, 2026

maarteNNNN commented on Feb 2, 2026

moonray commented on Feb 2, 2026

rekram1-node commented on Feb 3, 2026

rekram1-node commented on Feb 3, 2026

diegohb commented on Feb 3, 2026

LeoNardo-LB commented on Feb 3, 2026

TotalLag commented on Feb 4, 2026

sageil commented on Feb 17, 2026

KyrillGobber commented on Feb 25, 2026

sageil commented on Feb 25, 2026

moonray commented on Feb 25, 2026

sageil commented on Feb 26, 2026

github-actions commented on Jul 6, 2026

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions