BerriAI/litellm · error · HTTPException

soft_budget ({data.soft_budget}) must be strictly lower than

Error message

soft_budget ({data.soft_budget}) must be strictly lower than max_budget ({data.max_budget})

What it means

LiteLLM enforces that a project's soft_budget (alert threshold) is strictly lower than its max_budget (hard cap). If both are provided and soft_budget >= max_budget, the request is rejected with HTTP 400 before any DB write, because an alert threshold at or above the hard limit could never fire usefully.

Source

Thrown at enterprise/litellm_enterprise/proxy/management_endpoints/project_endpoints.py:131

    - Budget values are non-negative
    - soft_budget < max_budget
    """
    # --- Budget non-negativity checks ---
    if data.max_budget is not None and data.max_budget < 0:
        raise HTTPException(
            status_code=400,
            detail={"error": f"max_budget cannot be negative. Received: {data.max_budget}"},
        )
    if data.soft_budget is not None and data.soft_budget < 0:
        raise HTTPException(
            status_code=400,
            detail={"error": f"soft_budget cannot be negative. Received: {data.soft_budget}"},
        )

    # --- soft_budget < max_budget ---
    if data.soft_budget is not None and data.max_budget is not None:
        if data.soft_budget >= data.max_budget:
            raise HTTPException(
                status_code=400,
                detail={
                    "error": f"soft_budget ({data.soft_budget}) must be strictly lower than max_budget ({data.max_budget})"
                },
            )

    # --- Validate project models are a subset of team models ---
    project_models = data.models
    team_models = team_object.models or []
    if project_models and len(team_models) > 0:
        # If team has 'all-proxy-models', skip validation as it allows all models
        if SpecialModelNames.all_proxy_models.value not in team_models:
            for m in project_models:
                if m not in team_models:
                    raise HTTPException(
                        status_code=400,
                        detail={
                            "error": f"Model '{m}' not in team's allowed models. Team allowed models={team_models}. Team: {team_object.team_id}"

View on GitHub (pinned to 6c2dcb801b)

Solutions

Set soft_budget strictly below max_budget (e.g. 80% of max_budget)
If you swapped the fields, exchange the values and retry
Omit soft_budget (null) if you do not want threshold alerts

Example fix

// before
{"max_budget": 100, "soft_budget": 100}
// after
{"max_budget": 100, "soft_budget": 80}

Defensive patterns

Strategy: validation

Validate before calling

def check_budgets(p):
    sb, mb = p.get('soft_budget'), p.get('max_budget')
    if sb is not None and mb is not None and sb >= mb:
        p['soft_budget'] = round(mb * 0.8, 2)  # or raise ValueError locally

Type guard

const budgetsConsistent = (p) =>
  p.soft_budget == null || p.max_budget == null || p.soft_budget < p.max_budget;

Try / catch

catch (e) {
  if (e.status === 400 && /soft_budget.*strictly lower/.test(e.body?.detail?.error ?? '')) {
    body.soft_budget = body.max_budget * 0.8; return retry(body);
  }
  throw e;
}

Prevention

Always derive soft_budget as a fraction (e.g. 80%) of max_budget instead of hardcoding both
Add a pre-flight assert soft < max in CI config validation

When it happens

Trigger: POST /project/new or PUT /project/update with both soft_budget and max_budget set where soft_budget >= max_budget, e.g. soft_budget=100 and max_budget=100 (equality also fails; the check is >=).

Common situations: Setting the two budgets to the same round number, swapping the fields by mistake, or scaling one budget without scaling the other during a config migration.

Related errors

AI-assisted analysis of BerriAI/litellm@6c2dcb801b (2026-08-15). Data as JSON: /api/errors/8c1bd03028f68c03. Report an issue: GitHub.