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 has unveiled a multimodal framework that uses spatial tokens as the unit of reasoning.
Following the release of DeepSeek V4, major Chinese tech companies are engaged in fierce competition to secure Huawei AI chips.
As conflict between the US and China over AI technology theft and investment regulations intensifies, China's DeepSeek is narrowing the gap with low-cost, high-efficiency models.
In 8x H100 experiments, DeepSeek V4-Flash's mHC was found to become redundant after 3 layers and concentrate magnitude solely on BOS.
DeepSeek has teased the release of a new vision model.
DeepSeek cut V4-Pro pricing by 75% and also lowered API cache-hit costs to one-tenth of the previous rate.
The proliferation of open-weight models is shaking the monopoly-moat logic of US AI.
DeepSeek has lowered its API prices by up to 90% and also cut V4 costs by 75%.
Explains why the MoE architecture lowers inference costs for DeepSeek's large-scale model.
The US State Department has instructed diplomats worldwide to publicize allegations of IP theft by Chinese AI companies including DeepSeek.
New pricing policies from OpenAI and DeepSeek have polarized the AI cost structure.
Experimental inference support for DeepSeek v4 Flash has been released in llama.cpp.
DeepSeek V4's FP16 KV cache usage was recalculated based on a 1M context.
DeepSeek applied a limited-time 75% price cut to the V4 API.
The US State Department has issued a global warning over allegations that Chinese companies, including DeepSeek, are stealing AI technology.
DeepSeek has unveiled the V4 model, leveraging Huawei chips to boost performance while dramatically lowering prices.
DeepSeek released a preview of V4, emphasizing ultra-low-cost pricing and Huawei Ascend support.
DeepSeek-V4 was released, featuring 1M-token context and strong coding benchmark results.
DeepSeek has unveiled two variant models, **V4-Pro** and **V4-Flash**.
DeepSeek's latest models, the V4 Flash and Non-Flash versions, have been released via Hugging Face.
DeepSeek has been discussing an investment from Tencent and Alibaba at a valuation of $20 billion or more.
NVIDIA CEO Jensen Huang warned that U.S. export restrictions on AI chips to China could instead accelerate China's technological self-sufficiency.
Anthropic's release of Claude Opus 4.7 and OpenAI's $100 Pro launch, along with a ChatGPT outage, drew attention.