docs: add OpenUI + Ollama local setup tutorial with troubleshooting(issue #1)

[shogun444]

Summary

This PR replaces the fixed model selection in route.ts with process.env.MODEL || 'gpt-5.4', allowing model overrides through environment configuration while keeping the existing default behavior intact.

---

Changes

• Updated 1 single line in route.ts to use process.env.MODEL and removed hardcoded model dependency

-const MODEL = "gpt-5.4";
+const MODEL = process.env.MODEL || "gpt-5.4";

• Enabled flexible model configuration across environments
• No breaking changes

here's the command :

docker run --rm -p 3000:3000 \
  -e OPENAI_BASE_URL=http://host.docker.internal:11434/v1 \
  -e OPENAI_API_KEY=ollama \
  -e MODEL=minimax-m2.7:cloud \
  openui-chat

---

Why This Change

OpenUI depends heavily on structured, instruction-following outputs.

In practice:

• Small local models (3B–8B) frequently ignore instructions, hallucinate UI structure, and produce inconsistent results
• Cloud models generate stable and correct UI output

Relying purely on local models is not practical without modifying OpenUI internals or the prompt.

This change allows developers to:

• Use local models for experimentation
• Switch to reliable cloud models when needed
• Avoid editing core files just to change models

---

Usage

Run with Docker (set model via MODEL):

docker run --rm -p 3000:3000 -e OPENAI_BASE_URL=http://host.docker.internal:11434/v1 -e OPENAI_API_KEY=ollama -e MODEL=minimax-m2.7:cloud openui-chat 

Or set the model manually:

MODEL=minimax-m2.7:cloud

Then run OpenUI as usual.

---

Testing

• Verified working end-to-end at http://localhost:3000
• Tested with minimax-m2.7:cloud (recommended)
• Local models tested but found unreliable for UI generation

Proof

UI breaking in small models

Results After using the cloud models

---

Notes

• Focuses on minimal setup and developer flexibility
• Keeps Ollama as the local runtime
• Allows seamless switching between local and cloud models