Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.getunbound.ai/llms.txt

Use this file to discover all available pages before exploring further.

​
“Model Not Supported” (400 Error)

Why this happens: The model you’re trying to use hasn’t been enabled for your organization or tool setup. How to fix:
1

Log in to the dashboard

Go to your Unbound dashboard.
2

Open the Models page

Click Models in the left sidebar.
3

Check model availability

Verify the model you want is listed and available for your organization.
4

Contact your admin

If the model is listed but you still get the error, contact your organization administrator — they may need to enable it in your configuration.
Only organization administrators can enable or disable models. If you’re a regular member, ask your admin to make changes.

​
How to Switch Models

Different tools have different model switching mechanisms:
  • Unbound Chat — Use the model selector dropdown at the top of the chat interface
  • Claude Code CLI — The model is determined by your Unbound configuration. To request a different default model, contact your admin
  • Roo Code — Open Roo Code Settings (Ctrl/Cmd + Shift + P → “Roo Code: Open Settings”) and select a different model from the dropdown
  • Cursor — Model selection is configured within Cursor’s AI settings panel

​
Requesting New Models (e.g., Opus 4.6, GPT-5)

New models must be:
  1. Added to Unbound’s platform by the Unbound team (if not yet supported)
  2. Enabled for your organization by your admin
If a model you need isn’t available in the Models page:
  • Contact support to check if it’s on the roadmap
  • If it’s a recently released model, it may take a few days to be added

​
Token Limit Exceeded Errors

​
”Claude’s Response Exceeded the Output Token Maximum”

This error occurs when Claude generates a response larger than the configured maximum output tokens.

​
Fix for Claude Code CLI Users

Set the environment variable to increase the limit: 
export CLAUDE_CODE_MAX_OUTPUT_TOKENS=64000
Add this to your shell profile (~/.bashrc or ~/.zshrc) and restart your terminal.

​
Fix for API Users

Include the max_tokens parameter in your API request with a higher value: 
response = client.messages.create(
    model="claude-opus-4-5-20250514",
    max_tokens=8192,  # Increase this value
    messages=[{"role": "user", "content": "Your prompt"}]
)

​
General Tips to Avoid Token Limits

  • Break large tasks into smaller prompts — instead of “rewrite the entire file”, ask for one function at a time
  • Ask for concise responses — add “be brief” or “concise output” to your prompts
  • For code generation — ask for one file at a time instead of entire projects
  • Use streaming if your tool supports it — this allows partial responses even if the total would exceed limits

​
Overloaded Errors (529)

Error: 529 {"type":"error","error":{"type":"overloaded_error","message":"Overloaded"}} This is a temporary upstream error from the AI provider (e.g., Anthropic). It means the model is experiencing high traffic. How to handle:
  • Wait a few minutes and retry — these are usually brief
  • Try a different model — switch to an alternative (e.g., from Claude Opus to Claude Sonnet)
  • If persistent, contact support

​
Still Having Issues?

Contact support with:
  • The exact error message (screenshot preferred)
  • The model name you’re trying to use
  • Which tool you’re using (Claude Code, Roo Code, Cursor, Chat, API)

Model Policies

Allow or restrict specific models for your organization

List Models

See which models are available in your gateway