From 046c29dcf1083f0ab89b080f7696e6d642a6bd17 Mon Sep 17 00:00:00 2001 From: Benjamin Michaelis Date: Sat, 25 Oct 2025 11:12:08 -0700 Subject: fix: update OpenAI API to use max_completion_tokens instead of max_tokens (#2000) * fix: update OpenAI API to use max_completion_tokens instead of max_tokens The OpenAI API has deprecated max_tokens in favor of max_completion_tokens for newer models. This change updates both text and image model calls. * feat: add support for max_completion_tokens in OpenAI inference configuration --- packages/shared/config.ts | 2 ++ 1 file changed, 2 insertions(+) (limited to 'packages/shared/config.ts') diff --git a/packages/shared/config.ts b/packages/shared/config.ts index 5343246d..8e7d0252 100644 --- a/packages/shared/config.ts +++ b/packages/shared/config.ts @@ -65,6 +65,7 @@ const allEnv = z.object({ EMBEDDING_TEXT_MODEL: z.string().default("text-embedding-3-small"), INFERENCE_CONTEXT_LENGTH: z.coerce.number().default(2048), INFERENCE_MAX_OUTPUT_TOKENS: z.coerce.number().default(2048), + INFERENCE_USE_MAX_COMPLETION_TOKENS: stringBool("false"), INFERENCE_SUPPORTS_STRUCTURED_OUTPUT: optionalStringBool(), INFERENCE_OUTPUT_SCHEMA: z .enum(["structured", "json", "plain"]) @@ -241,6 +242,7 @@ const serverConfigSchema = allEnv.transform((val, ctx) => { inferredTagLang: val.INFERENCE_LANG, contextLength: val.INFERENCE_CONTEXT_LENGTH, maxOutputTokens: val.INFERENCE_MAX_OUTPUT_TOKENS, + useMaxCompletionTokens: val.INFERENCE_USE_MAX_COMPLETION_TOKENS, outputSchema: val.INFERENCE_SUPPORTS_STRUCTURED_OUTPUT !== undefined ? val.INFERENCE_SUPPORTS_STRUCTURED_OUTPUT -- cgit v1.2.3-70-g09d2