The US has decided not to blacklist DeepSeek, a Chinese company, along with over 100 other firms identified as security risks. This decision reflects ongoing tensions between the US and China regarding technology and security concerns.

DeepSeek V4 Pro is available at 5% of the cost of Claude and offers a range of features including hash-anchored edits, a sticky prefix cache, and autonomous loops for production code. It has been used for various applications such as training dose-prediction models for radiotherapy and developing a financial research agent, with no associated charges.

The GitHub repository "huggingface/open-r1" provides a fully open reproduction of the DeepSeek-R1 model. It includes scripts for installation, training, evaluation, and data generation, aiming to enable users to reproduce and build upon the R1 pipeline.

Posted by vinhnx. Score: 86 points. Comments: 63.

DeepSeek V4 Pro achieved a precision score of 38.0, outperforming GPT-5.5 Pro, which scored 33.0. DeepSeek excelled in handling overlapping patterns in a python log redactor task by using a single regex and replacer, while GPT-5.5 Pro utilized multiple regexes, leading to less effective results.

DeepSeek-V4-Flash is being implemented on the AMD MI300X, which launched in December 2023 as AMD's competitor to NVIDIA's H100 and H200 AI accelerators. The MI300X aims to address the current compute shortage while building an inference cloud for high-volume AI tasks.

DeepSeek API pricing is based on the number of tokens processed, charging per 1 million tokens for both input and output. The models available include deepseek-v4-flash and deepseek-v4-pro, accessible via specified base URLs for OpenAI and Anthropic formats.

DeepSeek has released two preview models in its V4 series: DeepSeek-V4-Pro and DeepSeek-V4-Flash. The Pro model features 1.6 trillion total parameters with 49 billion active, while the Flash model has 284 billion total parameters and 13 billion active, both utilizing a 1 million token context Mixture of Experts architecture under the MIT license.

The DeepSeek API is compatible with OpenAI and Anthropic formats, allowing access through their respective SDKs. Users can apply for an API key and utilize models such as deepseek-v4-flash and deepseek-v4-pro, with deepseek-chat set to be deprecated in 2026.

Three AI laboratories—DeepSeek, Moonshot, and MiniMax—conducted industrial-scale campaigns to illicitly extract Claude's capabilities, generating over 16 million exchanges through approximately 24,000 fraudulent accounts. These labs employed a technique called "distillation" to train less capable models using Claude's outputs, violating terms of service and access restrictions.