Ollama provider shows incorrect model list

[scmdavid]

Bug Description

When selecting Ollama (local) in AI Settings, the model dropdown displays models from other providers, including:

• GPT-5.2
• Zai-GLM 4.7

The expected Ollama models, such as llama3.2, are not available for selection.

Steps to Reproduce

1. Build the application with Docker:

docker build --no-cache -t openflowkit:debug .

2. Run it:

docker run --rm -p 3046:3045 openflowkit:debug

3. Open http://localhost:3046.
4. Navigate to Settings > AI Settings.
5. Select Ollama (local).
6. Open the Model dropdown.

Actual Behavior

The dropdown shows models such as GPT-5.2 and Zai-GLM 4.7. llama3.2 is not available.

Expected Behavior

The Ollama model dropdown should show Ollama-compatible models, for example:

• llama3.2
• llama3.1
• qwen2.5-coder
• mistral
• gemma3

The default model should be llama3.2.

Environment

• macOS
• Docker image built with docker build --no-cache
• Application served through nginx on port 3046

Additional Evidence

The source configuration defines Ollama models separately:

ollama: [
  { id: 'llama3.2', translateKey: 'llama3.2' },
  { id: 'llama3.1', translateKey: 'llama3.1' },
  { id: 'qwen2.5-coder', translateKey: 'qwen2.5-coder' },
  { id: 'mistral', translateKey: 'mistral' },
  { id: 'gemma3', translateKey: 'gemma3' },
]

However, the built application still renders models belonging to other providers. The Ollama model translations also appear to be missing from the English translation file, which may contribute to the issue.