Dangerous enough
When open source models claim to reach this accuracy levels in Biology, on top of that claiming to Nock Out all the frontier AI models, they are dangerous enough to...
Hugston.com News
Publish articleNews, analysis, and member publications
Qwen3.8-2.4T-A95B Open source claims to beat Fable 5 and Opus 4.8!
Qwen3.8-2.4T-A95B Open source claims to be better than Fable 5 and Opus 4.8 in Swe and coding.
Qwen 35B AgentWorld insights
Qwen-AgentWorld, the first language world model that simulates seven agent environments through long chain-of-thought reasoning: MCP, Search, Terminal, Software Engineering, Android, Web, and OS.
Anatomy of Z AI Kubernetes pod running on Kata Containers
Kubernetes pod running on Kata Containers (a hardware-level microVM)
Anatomy of Chatgpt Agents container and...
This is a small taste of how the trillion dollar business is done.
Glm 5.2 is out open weights
Z AI it released GLM-5.2
Ban European account, pay billions
The end of the digital monopoly
Local AI model claim to beat GPT 5.5 and Opus 4.7
Can a 400b parameters AI model beat GPT 5.5 and Opus 4.7?
The subscription to incompetence, is low IQ funding AI
Is there a correlation between Low IQ and AI expenditure?
Microsoft MAI Thinking1 and MAI Family models
MAI-Thinking-1 as a 35B active parameter MoE with a 256K ctx
Loading and running every AI llm Model with 4 Million ctx
Pushing the frontier ai models to their limit
HugstonOne 1 Million Dollar Edition
The real question is are you eligible?
Deepseek and OpenAI "integrity"
Whats your Integrity worth?
HugstonOne migrates to Codeberg (Europe)
Migrating to Europe
The west strategic battle defeat that may cost the entire AI race war
The rival hidden strategy to win.
DeepSeek-V4-Flash and DeepSeek-V4-Pro: The Million Token open weights
DeepSeek-V4-Flash 284B 13B parameters 1M ctx, DeepSeek-V4-Pro 1.6T parameters 49B active 1M ctx