Wednesday, July 29, 2026
Today's brief explores cutting local LLM memory requirements to just 2GB and optimizing production API costs through dynamic multi-model routing.
In this issue · 7
The Model Context Protocol C# SDK reached v2.0, bringing full compliance with the 2026-07-28 spec revision. The HTTP transport is now stateless by default, eliminating session handshakes and sticky routing requirements for horizontal deployments.
Liquid AI released two open-weight bidirectional encoder models, LFM2.5-Encoder-230M and 350M, optimized for long context. On CPU, the 230M model processes 8,192 tokens in 28 seconds—3.7x faster than ModernBERT-base.
OpenAI released `@openai/codex-security`, a command-line tool and TypeScript SDK designed to scan codebases, track security findings, and integrate automated security reviews into Continuous Integration pipelines.
Anthropic demonstrated that Claude Mythos Preview can autonomously discover mathematical flaws in core cryptographic algorithms. The model halved the key strength of the post-quantum candidate HAWK in 60 hours and accelerated round-reduced Advanced Encryption Standard attacks by up to 800 times. Anthropic also released CryptanalysisBench in collaboration with academic institutions.
An OpenAI study across eight scientific computing projects reveals how Codex and Claude Code achieved up to 60x speedups, 31% runtime cuts, and successful Rust/PyTorch migrations. The report highlights that human verification remains the core bottleneck.
TurboFieldfare is a custom Swift and Metal runtime that runs the 26-billion-parameter Gemma 4 model on Macs with limited RAM. By streaming experts from SSD instead of loading the entire model, it achieves efficient inference on 8 GB Macs.
Tokenless is an API router that queries multiple LLMs in parallel and monitors early output progress. Once a cheaper model shows high-quality compliance on a task, other models are cancelled, saving significant costs.
Email digest
One email a day — the stories that matter for engineers, founders and tech leads. Human-edited, with links to primary sources.
By subscribing you agree to the privacy policy.