[AI] Ollama Cloud: GLM 5.1 vs Gemma - How I got a second opinion between coding agents.

[Mashiane]

Hi Fam

So Im investigating why some users intellisense is sometimes not working on the VS B4X IDE Companion.

I have ran the same command on vs code.

GLM

ollama launch claude --model glm-5.1:cloud

Gemma

ollama launch claude --model gemma4

Prompt:

There is another report of one user not getting intellisense working, lets investigate what could be the potential reasons, just detail your finding, no need to implement any changes. Going forward any changes will have to be approved for implementation. Use debug skills, brainstorming and vscode extension building skills and other relevant skills.

GLMs Assestment of findings of the potential bug from both agents vs Gemma.

How gemma responded is more through text rather than tables here, however...

Gemma: The other agent's report is a treasure map of bugs, and my report is the toolset for verification. The other agent identified what is broken; I identified how to prove it's broken in the wild via the b4x.checkHealth command and specific instrumentation points.

I was under the impression that the "coding agent" had something to do with how the coding agent responds, it turns out thats not the case, its the internals of the coding models that determine what they output and how.

Why I chose GLM?

In terms of LLMArena, GLM is standing at #27 for text-text-task and #9 at coding.