From e98c437a72ade63f3117409ecad276b4ccd64c8b Mon Sep 17 00:00:00 2001 From: "github-actions[bot]" <41898282+github-actions[bot]@users.noreply.github.com> Date: Fri, 24 Jul 2026 10:54:28 +0000 Subject: [PATCH] docs: add models.default-ai-credits-pricing to glossary Adds a new glossary entry for the `models.default-ai-credits-pricing` frontmatter field introduced in #47687. This field allows workflow authors to provide fallback per-token pricing for BYOK/self-hosted models not in the AWF built-in pricing table, preventing HTTP 400 errors. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> --- docs/src/content/docs/reference/glossary.md | 11 +++++++++++ 1 file changed, 11 insertions(+) diff --git a/docs/src/content/docs/reference/glossary.md b/docs/src/content/docs/reference/glossary.md index 322de80c99b..370e5f6a4f0 100644 --- a/docs/src/content/docs/reference/glossary.md +++ b/docs/src/content/docs/reference/glossary.md @@ -776,6 +776,17 @@ models: blocked: ["*-preview"] ``` +### Default AI Credits Pricing (`models.default-ai-credits-pricing`) + +A `models:` frontmatter field that provides a fallback per-token pricing rate (in $/1M tokens) for models not present in the AWF built-in pricing table. When the AWF API proxy encounters an unrecognized model — such as a self-hosted BYOK Ollama or vLLM instance — it normally rejects the request with HTTP 400 `unknown_model_ai_credits`. Setting `default-ai-credits-pricing` supplies `input` and `output` rates that the proxy uses instead, preventing the rejection. Set both to `0` for free or local models. + +```aw wrap +models: + default-ai-credits-pricing: + input: 0 # $/1M input tokens + output: 0 # $/1M output tokens +``` + ### Max AI Credits (`max-ai-credits`) A top-level frontmatter field that caps the total AI Credits (AIC) the AWF proxy will spend within a single workflow run. Applies to all engines and maps to `apiProxy.maxAiCredits` in the compiled lock file. Defaults to `1000` when omitted. Accepts an integer, an optional `K`/`M` suffix string (for example, `100M`), or a GitHub Actions expression that resolves to an integer at runtime. Example: