abi/screenshot-to-code · warning · HTTPException

At least one model is required

Error message

At least one model is required

What it means

400 raised by POST /run_evals_stream in backend/routes/evals.py:267 when the request body's models list is empty. At least one model must be selected to schedule evaluation tasks; the guard fires before any set/session resolution.

Source

Thrown at backend/routes/evals.py:267

        if selected_files:
            return len([f for f in selected_files if f in brief_ids])
        return len(brief_ids)

    if selected_files:
        return len([f for f in selected_files if f.endswith(".png")])

    if set_name:
        return len(eval_sets.list_set_images(set_name))

    input_dir = os.path.join(EVALS_DIR, "inputs")
    return len([f for f in os.listdir(input_dir) if f.endswith(".png")])


@router.post("/run_evals_stream")
async def run_evals_stream(request: RunEvalsRequest):
    """Run evaluations and stream progress events as newline-delimited JSON."""
    if not request.models:
        raise HTTPException(status_code=400, detail="At least one model is required")

    eval_set, session, skip_per_model = _resolve_set_run(request)

    per_model_task_counts: Dict[str, int] = {}
    per_model_skipped_existing: Dict[str, int] = {}
    if request.diff_mode:
        for model in request.models:
            pending_tasks, skipped_tasks = count_pending_eval_tasks(
                stack=request.stack,
                model=model,
                input_files=request.files,
                n=N,
                diff_mode=True,
                eval_set=eval_set,
                skip_input_files=skip_per_model.get(model),
            )
            per_model_task_counts[model] = pending_tasks
            per_model_skipped_existing[model] = skipped_tasks

View on GitHub (pinned to d026163f58)

Solutions

1. Include at least one valid model name in the models array.
2. Disable the run button in the UI until at least one model is selected.
3. If no models appear selectable, check that an API key is configured (see the models the backend offers).

Example fix

// before
if (runRequested) startRun({ models: selectedModels, ... });

// after
if (selectedModels.length === 0) { toast('Select at least one model'); return; }
startRun({ models: selectedModels, ... });

if (!Array.isArray(models) || models.length === 0) {
  throw new Error('At least one model is required');
}

function hasModels(models: unknown): models is string[] {
  return Array.isArray(models) && models.length > 0 && models.every(m => typeof m === 'string');
}

When it happens

Trigger: POST /run_evals_stream with models: [] — e.g. the UI sent the request before any model checkbox was ticked, or a filter removed all models.

Common situations: Default model list empty due to missing API keys filtering out all models; race where the run button is enabled before model selection loads.

Related errors

• Invalid run id
• max_age_days must be >= 1
• Design system name is required
• Invalid eval set name: {set_name!r}
• Not a set image: {filename!r}

AI-assisted analysis of abi/screenshot-to-code@d026163f58 (2026-08-14). Data as JSON: /api/errors/7596224e5ea6d9d1. Report an issue: GitHub.