DeepSeek and Alibaba cut prices as n8n, Langflow patch security bugs
46,973 stars today78 new projects+11,354 rea
By olud.ai editorial · built from our own tracking data, published daily
DeepSeek and Alibaba led a broad price reduction wave on October 21, with seven models seeing input price drops ranging from 5% to 33%. DeepSeek V4 Pro 0423 fell 5% to $0.35 per 1M tokens, while Qwen3.6 35B A3B dropped 33% to $0.10 per 1M tokens. Two models bucked the trend: Qwen3 Next 80B A3B Instruct rose 11% and Solar Pro 4 jumped 67%.
The open-source AI ecosystem added 46,973 GitHub stars today across the 10,000+ tracked projects, with 78 new projects entering the index. The weekly total stands at 2,885 new projects and 100 new Hugging Face spaces. The top gainer "rea" pulled in +11,354 stars.
Releases of the day
Five minor releases shipped today, all focused on bug fixes and security patches. n8n 2.44.5 fixes key queue job outcomes on shutdown resume and loads credential types on startup. Langflow v1.12.7 patches OpenSearch connector URL validation and other security scoping issues. MLflow 3.17.2 adds evaluation lineage tracking for datasets, scorers, and agents. Pydantic AI v2.59.0 addresses a high-severity HTML conversion vulnerability in web_fetch. Stability Matrix v2.16.5 introduces Fizgig, a LoRA training studio for Flux 2 Klein 9B, Krea 2, and MiniMax models.
Seven models saw input price decreases today. Alibaba led with Qwen3.6 35B A3B down 33% to $0.10, Qwen3.5 397B A17B down 22% to $0.39, and Qwen3.5-35B-A3B down 7% to $0.14. DeepSeek V3 dropped 19% to $0.26, Gemma 3 4B fell 20% to $0.04, and DeepSeek V4 Pro 0423 slipped 5% to $0.35. Two models increased: Qwen3 Next 80B A3B Instruct rose 11% to $0.10, and Solar Pro 4 jumped 67% to $0.15.
▼ 5%DeepSeek V4 Pro 0423DeepSeek · input $0.37 → $0.35 per 1M tokens
▲ 11%Qwen3 Next 80B A3B InstructAlibaba · input $0.09 → $0.1 per 1M tokens
▲ 67%Solar Pro 4Upstage · input $0.09 → $0.15 per 1M tokens
Research pick
A Hugging Face community paper titled "Focused Omni-Modal GUI Agents" argues that current GUI agents waste resources by processing all available streaming data at every step. The authors propose selective attention mechanisms to reduce computational load while maintaining task performance.
The simultaneous price cuts from DeepSeek and Alibaba signal intensifying competition in the model inference market, while the flurry of security patches across n8n, Langflow, and Pydantic AI highlights growing operational risk in AI toolchains.
Source: olud.ai tracking of 10,000+ open-source AI projects, 300+ models and live provider pricing. Figures are measured, not estimated. All releases · Live pricing · Latest in AI
