[Bug]: Issue using litellm with pi.dev

Check for existing issues

• I have searched the existing issues and checked that my issue is not a duplicate.

What happened?

When I use pi.dev with models in ollama it works as expected. But when i bring in litellm to talk with ollama I see weird tool call output from same ollama model.

On using ollama models directly as shown below I am seeing right output from respective models.

cat models.json
{
  "providers": {
    "ollama": {
      "baseUrl": "http://10.30.0.103:11434/v1",
      "api": "openai-completions",
      "apiKey": "ollama",
      "models": [
        { "id": "gemma4:e4b" },
        { "id": "qwen3.5:9b" }
      ]
    }
  }
}

 pi v0.74.0
 escape interrupt · ctrl+c/ctrl+d clear/exit · / commands · ! bash · ctrl+o more
 Press ctrl+o to show full startup help and loaded resources.

 Pi can explain its own features and look up its docs. Ask it how to use or extend Pi.


 say hello


 The user requested a simple greeting: "say hello". I should respond with a friendly, textual greeting.

 Hello! How can I help you today?

────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────

Only showing models from configured providers. Use /login to add providers.

>

→ gemma4:e4b [ollama] ✓
  qwen3.5:9b [ollama]

  Model Name: gemma4:e4b

Now I am using below configuration to use litellm as proxy to talk with same ollama models. I am seeing output in a structured format instead of final output in case of ollama models

cat models.json
{
  "providers": {
    "litellm": {
      "baseUrl": "http://10.30.0.103:4000/v1",
      "api": "openai-completions",
      "apiKey": "sk-1234",
      "compat": {
        "supportsDeveloperRole": false,
        "supportsReasoningEffort": false
      },
      "models": [
        {
          "id": "ollama/qwen3.5:9b",
          "reasoning": false
        },
        {
          "id": "ollama/gemma4:e4b",
          "reasoning": false
        }
      ]
    }
  }
}
pi

 pi v0.74.0
 escape interrupt · ctrl+c/ctrl+d clear/exit · / commands · ! bash · ctrl+o more
 Press ctrl+o to show full startup help and loaded resources.

 Pi can explain its own features and look up its docs. Ask it how to use or extend Pi.


 say hello


 {
   "name": "say_hello",
   "arguments": {
     "text": "Hello!"
   }
 }

────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────

Only showing models from configured providers. Use /login to add providers.

>

→ ollama/qwen3.5:9b [litellm] ✓
  ollama/gemma4:e4b [litellm]

  Model Name: ollama/qwen3.5:9b

Steps to Reproduce

1. Install pi.dev and configure models.json as mentioned in description
2. run pi command to start the terminal

Relevant log output

All logs are attached in description.

What part of LiteLLM is this about?

Proxy

What LiteLLM version are you on ?

v1.83.10

Twitter / LinkedIn details

No response