Hmbown/CodeWhale · error · anyhow::Error

maxOutputTokens requires an exact model

Error message

maxOutputTokens requires an exact model

What it means

Setting `maxOutputTokens` requires resolving an exact model identifier: the caller's requested model, or the provider's `default_model`. If neither exists, or the resolved value is empty/`auto`, the app-server cannot look up the model's output-token capability and refuses. `auto` is not a concrete model, so no catalog lookup is possible.

Solutions

1. Pass an explicit model with the request (e.g. `model: "gpt-5"` style identifier accepted by the provider).
2. Set the provider's `default_model` in the Runtime configuration.
3. Drop maxOutputTokens if automatic model selection is intended.

Example fix

// before
{ "prompt": "hi", "maxOutputTokens": 1024 }
// after
{ "prompt": "hi", "model": "qwen3-coder", "maxOutputTokens": 1024 }

// client-side guard before sending maxOutputTokens
function canSendMaxOutputTokens(model, provider) {
  const resolved = model ?? provider?.default_model;
  return typeof resolved === 'string' && resolved.trim() !== '' && resolved.toLowerCase() !== 'auto';
}

function isExactModel(v) {
  return typeof v === 'string' && v.trim().length > 0 && v.toLowerCase() !== 'auto';
}

When it happens

Trigger: Calling the one-shot prompt path with `max_output_tokens` set while the requested model is None and the Runtime provider's `default_model` is absent, empty, or literally "auto".

Common situations: Clients passing maxOutputTokens without pinning a model while the runtime provider is left on automatic model selection.

Understand the failure class

Background: "missing required argument" and "the following required arguments were not provided": what required-argument errors mean and how to fix them — this error's family across 20 libraries.

Related errors

• A pinned task provider requires an explicit model
• agent profile reasoning_effort must be one of: inherit…
• At least one thread field is required
• Choose both a provider and a model.
• Codewhale credentials path must be a directory

AI-assisted analysis of Hmbown/CodeWhale@73e0f67d83 (2026-09-22). Data as JSON: /api/errors/3b2d514643ee4480. Report an issue: GitHub.

Appendix: source

Thrown at crates/app-server/src/lib.rs:1651

            .await?;
        let current = providers
            .get("current")
            .and_then(Value::as_str)
            .context("Runtime provider is unavailable")?;
        let provider = providers
            .get("providers")
            .and_then(Value::as_array)
            .and_then(|providers| {
                providers
                    .iter()
                    .find(|provider| provider.get("id").and_then(Value::as_str) == Some(current))
            })
            .context("Runtime provider is unavailable")?;
        let model = requested_model
            .or_else(|| provider.get("default_model").and_then(Value::as_str))
            .context("maxOutputTokens requires an exact model")?;
        if model.trim().is_empty() || model.eq_ignore_ascii_case("auto") {
            bail!("maxOutputTokens requires an exact model");
        }
        if !current
            .bytes()
            .all(|byte| byte.is_ascii_lowercase() || byte.is_ascii_digit() || byte == b'-')
        {
            bail!("Runtime provider identity is invalid");
        }
        let mut cursor = None;
        let mut seen = std::collections::HashSet::new();
        loop {
            let mut url =
                reqwest::Url::parse(&format!("{}/v1/providers/{current}/models", self.base_url))?;
            url.query_pairs_mut().append_pair("limit", "250");
            if let Some(cursor) = cursor.as_deref() {
                url.query_pairs_mut().append_pair("cursor", cursor);
            }
            let catalog = self.request_json(self.authed(self.client.get(url))).await?;
            if let Some(entry) =

View on GitHub (pinned to 73e0f67d83)