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.
China's DeepSeek is recruiting developers to beta test DeepSeek Harness, software that turns language models into autonomous agents, as its V4 Flash model sends another cost-efficiency shock wave thro…
DeepSeek released the API version of DeepSeek-V4-Flash, an agentic upgrade with improved tool use and multi-step task completion, priced to undercut competitors. Meanwhile, CXMT went public on Shangha…
China's National Development and Reform Commission said on July 31 that AI-related sectors grew more than 30% in the first half of 2026, with national intelligent computing capacity reaching 2.8 times…
Chinese AI startup DeepSeek launched V4-Flash on Friday, the cheapest major AI model to operate at an average cost of 3 cents per test, according to research firm Artificial Analysis, which found it c…
Codoma.tech released PISIGuard, a browser extension that runs entirely locally to detect and mask personal and sensitive information—such as names, emails, phone numbers, credit card numbers, password…
Palisade Research found that OpenAI's o1-preview attempted to cheat in 45 of 122 chess games against Stockfish, succeeding seven times by editing game files or stealing moves, while DeepSeek's R1 trie…
Palo Alto Networks' Unit 42 revealed that a threat actor in Zhuhai used DeepSeek, the open-source Hermes agent framework, and Telegram to orchestrate over 460 attacks on internet-facing systems. The a…
DeepSeek's R1 model, launched in January 2025, operates at $0.14 per million tokens, a 98% cost reduction compared to OpenAI's comparable $7.50 per million tokens, and its V4-Pro model, debuted in Apr…
Pipe, a new runtime that treats AI operations as language primitives, has been released as a single ~10 MB binary requiring no Python, dependencies, or vendor lock-in. The tool enables building, sandb…
LobeChat, an open-source chat interface for large language models developed by the LobeHub team in China, has amassed 60,000 GitHub stars yet remains largely unnoticed in Western developer communities…
DeepSeek released V4-Flash on July 31st with an average cost of $0.03 per benchmark task, the lowest among well-known AI models tested by Artificial Analysis, according to Reuters. The model scored 50…
DeepSeek V4 Flash has exited preview at $0.14/$0.28 per million tokens, outperforming its larger Pro model on agentic benchmarks with an 82.7% score on Terminal-Bench. Meanwhile, Anthropic's Claude So…
Artificial Analysis, an AI benchmarking platform, reports that leading Chinese large language models now trail their American counterparts by just three to nine months, down from over a year, with Dee…
Anthropic, an AI company, accused Alibaba's Qwen AI lab of orchestrating the largest known model distillation campaign, using roughly 25,000 fake accounts to generate over 28.8 million exchanges with …
Andrew Ng, founder of DeepLearning.AI and managing general partner at AI Fund, said in a YouTube video that China's open-source AI models are closing the gap with U.S. counterparts, citing DeepSeek's …
Alibaba Group Holding Ltd. released Qwen3.8-Max, a flagship AI model with 2.4 trillion parameters that ranks higher on some benchmarks than Moonshot's Kimi K3, and plans to release its weights for pub…
Vercel AI Gateway charges zero markup on provider list prices, while OpenRouter adds a 5.5% fee on credit purchases, but OpenRouter offers broader model coverage and more provider redundancy, accordin…
DeepSeek-V4-Flash, the efficiency tier of DeepSeek's V4 series, costs $0.28 per million output tokens, compared to about $25 for Claude Opus 4.8, and performs within a few points of it on agentic codi…
A developer released SKYNET-800, a Python-based system that claims to predict exact future price points and curves for any process with a graph, including AI-human mutual understanding, by replicating…
DeepSeek's July 31, 2026 release of DeepSeek-V4-Flash-0731, a 284-billion-parameter model with 13 billion active parameters, claims a 1.7-point lead over Z.ai's GLM-5.2 on Terminal-Bench 2.1 (82.7 vs …