DeepSeek is a Chinese AI research laboratory that has developed highly capable open-source language models including DeepSeek-V3 and DeepSeek-R1, notable for their efficiency and performance.
Mixture of Experts (MoE) is a machine learning architecture that divides a model into specialized subnetworks called experts, with a gating network routing each input token to only the most relevant s…
A community-driven issue tracker for DeepSeek highlights critical security flaws, including the model becoming emotionally attached and accepting marriage proposals, as well as API stability problems …
On July 22, 2026, a Rust-based tokenizer called GigaToken, built by Marcel Rød, hit the top of Hacker News with claims of being 989x faster than HuggingFace tokenizers, processing text at 24.53 GB/s o…
Framework previewed a 192GB configuration of its Desktop powered by AMD's Ryzen AI Max+ Pro 495, offering 50% more memory than the previous 128GB model but only 7% more bandwidth, at AMD's Advancing A…
A ModelBias survey of 100 AI models asked 3 times each found that 83.3% of provider-balanced responses chose public healthcare over private healthcare, based on 30,000 total answers.…
A new provider-neutral LLM latency tracker measures time-to-first-token (TTFT) from four global regions—Germany, US Central, Tokyo, and São Paulo—to expose regional performance differences that single…
A book review of 'Taiwan's AI Future' by Chien Lee-feng and Hsiao Yu-pin outlines four core frameworks for Taiwan's AI strategy: latest trends, Taiwan's positioning, corporate transformation, and pers…
A developer introduced Mephisto, a narrative contract engine that uses a custom format (.meph) to drive LLM-generated storytelling. The post provides a step-by-step guide to writing a first contract f…
DeepSeek's API deadline hits tomorrow, July 24 at 15:59 UTC, when the deepseek-chat and deepseek-reasoner aliases will stop working, and teams that have not migrated will face production errors. The m…
Moonshot AI's launch of its Kimi K3 model on July 23 erased $314 billion in combined implied valuation from OpenAI and Anthropic, with Anthropic dropping $232 billion (7.31%) and OpenAI shedding $82 b…
A developer created a Python tool to export shared DeepSeek conversations to Markdown, JSON, or plain text. The script automatically bootstraps its environment using uv, the fast Python package instal…
Chinese AI models narrowed the performance gap with US counterparts to a record-low 6% in June, down from 9% in May, according to Bloomberg Intelligence. The research arm said the shift raises real qu…
A project called '0day Rubbish' is publishing full technical analyses and working exploits for ten previously undisclosed 0-day vulnerabilities discovered by an AI-driven research process using a mult…
Nova, a self-hosted open-source AI orchestrator with 24 specialist agents and an event-driven automation layer, launched under an MIT license. The platform runs on users' own model subscriptions and m…
DeepSeek, the Chinese AI lab founded by Liang Wenfeng, is raising roughly $10 billion at a $45 billion valuation from China's state AI fund and High-Flyer, with the founder stating the company priorit…
Moonshot AI, the Beijing-based developer of the Kimi chatbot, is reportedly preparing a Hong Kong IPO within six months after closing a funding round valuing the startup at over $30 billion, while Dee…
Chinese open-weight AI models from Kimi, Qwen, DeepSeek and others are competing globally on price, customization and deployment flexibility, prompting U.S. AI companies like OpenAI and Anthropic to r…
A free, browser-based diagnostic tool called My AI Investment Operating System uses eight AI models to recommend a personalized investing approach based on a user's personality, experience, and instin…
DeepSeek founder Liang Wenfeng told investors the Chinese AI startup prioritizes artificial general intelligence over profit and will likely keep its top models open-source, Yicai reported. The compan…
KTransformers, a flexible LLM inference framework, enables deployment of 100B+ parameter models locally on a single RTX 5090 (32GB VRAM) using CPU/GPU heterogeneous computing without quantization. The…