AI BriefingAlibaba raises DeepSeek V4 Flash 0731 output prices by 230%
·2026.10.20 09:00
Key point
Output costs for DeepSeek V4 Flash 0731 increased from $0.40 to $1.32 per million tokens, part of a broader wave of API price hikes across major labs.
Details
- Alibaba increased the output price of DeepSeek V4 Flash 0731 by 230%, raising costs from $0.40 to $1.32 per million tokens, while input costs rose from $0.20 to $0.44.
- Stork's price tracking identified 17 distinct price movements, including 150% and 50% hikes for Mistral's Ministral 3B and 8B models, and a 52% increase for Alibaba's Qwen3.6 35B-A3B.
- Zhipu AI was the only major provider to lower prices, slashing GLM-5V-Turbo by 85%.
- The price surge reflects the end of subsidized inference, forcing developers to optimize prompts and implement caching to maintain margins.
Related AI Developments
- NVIDIA launched Nemotron 3.5, a 120B parameter model with 12B active parameters per token, designed to reduce inference costs through hybrid Mamba-Transformer and MoE architecture.
- Meta released Muse Wonder Full and the 30B open-weight Muse Glimmer, the latter optimized for single consumer GPU deployment.
- MiniMax officially launched M3.1 Flash, confirming it as the stealth 'Space Bunny' model, which claims to run at one-third the cost and three times the speed of M3.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.
Related
DeepSeek open-sources DeepInfer inference framework for V4.1-Flash
Meta revises commercial terms to remove 'train' but retains broad scraping rights for AI business products