From 8a18b70afff5228e5cccf2be5147b4258b0ff1ca Mon Sep 17 00:00:00 2001
From: oabrivard <olivier@abrivard.fr>
Date: Mon, 23 Mar 2026 23:37:39 +0100
Subject: [PATCH] fix: set max output tokens to 16384 for all LLM providers

OpenAI's default output limit (4096 tokens) was too low for structured
synthesis output with multiple categories and articles per category,
causing truncated JSON. Set 16384 for both OpenAI APIs (Responses +
Chat Completions) and Gemini. Anthropic already had 16384.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
---
 backend/src/services/llm/gemini.rs | 3 ++-
 backend/src/services/llm/openai.rs | 2 ++
 2 files changed, 4 insertions(+), 1 deletion(-)

diff --git a/backend/src/services/llm/gemini.rs b/backend/src/services/llm/gemini.rs
index a4186f8..4f33cf4 100644
--- a/backend/src/services/llm/gemini.rs
+++ b/backend/src/services/llm/gemini.rs
@@ -138,7 +138,8 @@ fn build_request_body(
         },
         "generationConfig": {
             "responseMimeType": "application/json",
-            "responseSchema": response_schema
+            "responseSchema": response_schema,
+            "maxOutputTokens": 16384
         }
     });
 
diff --git a/backend/src/services/llm/openai.rs b/backend/src/services/llm/openai.rs
index 8d0dbfc..ea431be 100644
--- a/backend/src/services/llm/openai.rs
+++ b/backend/src/services/llm/openai.rs
@@ -44,6 +44,7 @@ impl OpenAiProvider {
             "model": model,
             "instructions": system_prompt,
             "input": user_prompt,
+            "max_output_tokens": 16384,
             "text": {
                 "format": {
                     "type": "json_schema",
@@ -116,6 +117,7 @@ impl OpenAiProvider {
                     "content": user_prompt
                 }
             ],
+            "max_tokens": 16384,
             "response_format": {
                 "type": "json_schema",
                 "json_schema": {