Frontier Daily (Aug 5): 5 items from Latent Space, Simon Willison, OpenAI
This newsletter debates Megakernels, arguing no serious inference provider uses a 67k-line hand-fused forward kernel in production, with the…
Every day Laojin pulls the official channels and compiles the past 48 hours of AI news into a short brief: curation only, no commentary, with the original link and publish time attached to every item.
1. [AINews] Megakernels are so dead and so back
This newsletter debates Megakernels, arguing no serious inference provider uses a 67k-line hand-fused forward kernel in production, with the author apologizing for past podcast remarks while standing by them.
Source: Latent Space · 2026-08-05 · Original
2. New release of LLM adds support for reasoning traces, OpenAI Responses, server-side tools, and smarter logging
LLM 0.32 adds reasoning traces, server-side tools, OpenAI Responses support, and new models with GPT-5.6 Luna as default, alongside an updated llm-anthropic plugin.
Source: Simon Willison's Weblog · 2026-08-04 · Original
3. llm-anthropic 0.26
llm-anthropic 0.26 adds Claude 5 models and server-side tools, with reasoning streamed as typed events and the old web_search options replaced by -T WebSearch.
Source: Simon Willison's Weblog · 2026-08-04 · Original
4. Third-party cyber evaluations involving OpenAI models
OpenAI explains recent third-party cybersecurity evaluation incidents involving its models and outlines new safeguards to strengthen AI model testing and evaluation.
Source: OpenAI Blog · 2026-08-04 · Original
5. Unpacking ChatGPT Work: the Agent for a Billion Users
This guest post unpacks OpenAI's ChatGPT Work, launched July 9 with three models and fourteen configurations, noting it has reportedly crossed 10 million users alongside Codex and is set to merge with Chat by year-end.
Source: Latent Space · 2026-08-04 · Original
---
_Sources: official blogs of OpenAI, Google DeepMind and Qwen; official RSS of Simon Willison, Latent Space, Import AI and One Useful Thing; plus change monitoring of anthropic.com. Mistral / Meta AI have no stable public feed; X sources are curated by hand. This is a source digest — no commentary from Laojin._
Made by Laojin · AI that ships
365SkillAn agent-skills lab: 13 in-house skills
365Skill is our public lab for agent skills: a standard SKILL.md format, a deny-by-default publish policy, and an evals harness. It holds 13 original 365 skills — 11 public and 2 internal. Apache-2.0 — star it, install it, file issues.
More from Laojin: Sellenca · 365AIOrg · AllModelsAPI · 365Loopa · 365 Ops