AI Briefing#deepseek
The latest AI and developer news about #deepseek, with the original source and a short summary.
The latest AI and developer news about #deepseek, with the original source and a short summary.
DeepSeek V4.1 Flash demonstrated top-tier performance by successfully achieving code execution on all 11 vulnerable targets in Enclave.ai's hacking benchmark.
DeepSeek released the 552B MoE model V4.1-Flash, replacing V4-Pro and optimizing KV cache to 890 bytes per token.
DeepSeek has released V4.1-Flash, which offers superior performance and efficiency compared to V4-Pro, along with reduced API prices.
DeepSeek V4.1 Flash is now available on AI Gateway, supporting native image understanding and a 1 million token context window.
DeepSeek announced that it will officially launch v4.1 Flash on September 10, stating it outperforms v4 Pro across all metrics.
Magic announced that a new pretraining recipe reached performance similar to DeepSeek V4 Pro Base with approximately 50x fewer FLOPs, and that training scaled up by approximately 10x subsequently outperformed public base models in perplexity evaluations.
Empirical data demonstrates that on OpenRouter, even for the same model, performance and features such as benchmark scores, vision recognition, and parameter support can vary significantly depending on the Provider.
NVIDIA has released the NVFP4 quantized version of DeepSeek-V4-Pro-0813, supporting a MoE architecture that activates 400B parameters out of a total of 1.65T.
The value of AI models varies according to the intelligence required for each task, with excessive intelligence leading to inefficiency.
DeepSeek has released the v4-flash vision API, which processes text and images together.
An experimental vision-capable version of DeepSeek V4 Flash has been released on Vercel AI Gateway.
DeepSeek has released V4-Flash-Vision-Exp, significantly enhancing multimodal agent performance while maintaining text capabilities.
LMSys has released optimization strategies to maximize DeepSeek-V4-Pro serving performance on H20 GPUs.
This analysis examines the diversification strategy of China's AI industry, focusing on the founding of Pragmatik Labs by an Alibaba alumnus and the competitiveness of Moonshot AI's open-weight models.
OpenAI and Zhipu have launched cyber-specialized models, followed by model updates from major companies such as Meta and Alibaba.
DeepSeek is implementing API price increases of up to 1,100% per model starting this Sunday.
Major companies and state-owned entities, including DeepSeek and Tencent, are participating in Unitree's IPO, evaluating robotics as an infrastructure industry.
DeepSeek Harness, released by DeepSeek AI, is a modular agent framework where even the agent loop can be replaced via plugins.
Chinese AI startups are evolving beyond merely chasing model performance, pursuing diverse paths such as AGI, open ecosystems, and commercialization.
DeepSeek introduces a differential API pricing model based on peak and off-peak hours alongside the launch of V4-Pro.
DeepSeek has released an open-source coding agent designed with plugins for models, tools, and the agent loop.
DeepSeek has released DeepSeek-V4-Pro with enhanced agent performance and revamped its API pricing plans.
DeepSeek has released the developer preview of DeepSeek Harness, a plugin-based framework for building agents.
AI Gateway now applies updated weights to DeepSeek V4 Pro by default.