4 min read
Claude Opus 4.8 on Azure now bills…
Anthropic brought Claude Opus 4.8 and Haiku 4.5 to Microsoft Foundry with Azure-native billing, Entra ID, and full prompt caching. Here is what enterprise teams should configure first.
Blog 385 posts
Notes from the trenches on AI engineering, LLM apps, and the full-stack work that holds it all together.
4 min read
Cursor for iOS turns your phone into…
Cursor shipped a native iOS app for launching cloud agents, steering local runs with Remote Control, and merging PRs from your phone. Here is how I use it without losing my local MCP stack.
4 min read
DeepSeek DSpark speeds LLM inference 60-85% without…
DeepSeek's DSpark speculative decoding framework adds a semi-autoregressive drafter and confidence-scheduled verification to V4 serving. Per-user generation runs 60-85% faster at matched throughput, lossless and open source.
4 min read
MegaTrain fits 120B LLM training on one…
MegaTrain stores weights in host memory and streams one layer at a time to the GPU, training 120B models on a single H200 with 1.5TB RAM. It beats DeepSpeed ZeRO-3 offload by 1.84x at 14B scale.
5 min read
Meta Brain2Qwerty v2 decodes full sentences from…
Brain2Qwerty v2 hits 78% word accuracy on the best participant using only a non-invasive MEG helmet. Meta open-sourced the training code. Here is what that means for applied AI and assistive tech.
4 min read
Ten open-source scraping tools that replace paid…
Firecrawl, browser-use, Crawl4AI, MarkItDown, and Crawlee can build training and RAG pipelines without $2,000/month scraper contracts. Here is the stack I actually wire for clients.
6 min read
How Coinbase cut AI spend 50% with…
Brian Armstrong's team runs 1,200 AI agents, defaults to GLM-5.2 and Kimi K2.7, and automated model selection. Five levers any engineering org can copy without a crypto-scale budget.
8 min read
GLM-5.2 is the first open-weight model that…
Z.ai's MIT-licensed GLM-5.2 ships 1M-token context, beats GPT-5.5 on several coding benches, and costs a fraction of Opus. Here's what production tests show, where it still breaks, and how I'd route it.