PulseAugur
EN
LIVE 15:10:14

Qwen4-Max-Preview token limit change causes empty replies

By PulseAugur Editorial · [1 sources] ·

A recent change in how Alibaba Group's Qwen4-Max-Preview model handles token limits has led to unexpected empty replies. Previously, `max_tokens` primarily capped the generated answer, but the model now uses this limit for both the reasoning process and the final output. This means the model can consume its entire token budget on thinking, leaving no room for an answer, even when the API returns a successful HTTP 200 status. Developers integrating with this model, particularly through platforms like AIHubMix, need to adjust their token limits or reasoning settings to accommodate this change, with Alibaba recommending the use of `max_completion_tokens` for new integrations. AI

IMPACT Developers using Qwen4-Max-Preview must adjust token settings to avoid empty responses due to changes in how reasoning and output tokens are counted.

RANK_REASON The item discusses a specific technical issue and its resolution for a particular model and platform, rather than a general release or significant industry event.

Read on dev.to — LLM tag →

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →

Qwen4-Max-Preview token limit change causes empty replies

How we ranked this

Signal score
54 / 100
Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket
Tool
The item discusses a specific technical issue and its resolution for a particular model and platform, rather than a general release or significant industry event.
Source corroboration
Single-source cluster
Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics
model release, product
Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance
High
Clearly on-topic for AI-industry coverage.
Story freshness
Breaking (< 6h)
Fresh story with cross-source coverage still developing. Ranking may shift as more sources report.

Full methodology in our editorial standards.

COVERAGE [1]

  1. dev.to — LLM tag TIER_1 English(EN) · AIHubMix ·

    Qwen4-Max-Preview returned an empty reply at max_tokens 300: here's why

    <p>A 300-token <code>max_tokens</code> used to be a safe cap for a short answer. On Qwen4-Max-Preview it returned HTTP 200 with an empty <code>content</code>. Alibaba changed what <code>max_tokens</code> covers: on this model it limits the thinking and the answer together, so the…

RELATED ENTITIES

RELATED TOPICS