From 2457a3a5ae2b64030c3bf9996f876b8b39ce6631 Mon Sep 17 00:00:00 2001 From: Vikhyat Korrapati Date: Fri, 26 Jun 2026 21:03:21 -0700 Subject: [PATCH] docs: use max_completion_tokens for the OpenAI-compatible endpoint OpenAI deprecated max_tokens in favor of max_completion_tokens; document the current parameter on the OpenAI Compatibility page. --- docs/openai.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/docs/openai.md b/docs/openai.md index 3859904..37d5446 100644 --- a/docs/openai.md +++ b/docs/openai.md @@ -184,7 +184,7 @@ See [Reasoning](/reasoning) for more on how Moondream 3 reasons. | `messages` | array | OpenAI chat messages. `content` may be a string or an array of `text` / `image_url` parts. | | `temperature` | number | Sampling temperature. | | `top_p` | number | Nucleus sampling. | -| `max_tokens` | integer | Maximum completion tokens (up to 4096). | +| `max_completion_tokens` | integer | Maximum number of tokens to generate, including reasoning tokens (up to 4096). | | `reasoning` | boolean | Enable the reasoning trace (returned on `message.reasoning`). | | `stream` | boolean | Stream the response as SSE. | | `stream_options.include_usage` | boolean | Emit a final usage chunk in a stream. |