Frontier Daily (Jul 25): 5 items from Anthropic, Simon Willison, Latent Space
Claude Opus 5 is a step change improvement for the Opus tier, powering long-running agents with better coding and professional work.
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. Introducing Claude Opus 5
Claude Opus 5 is a step change improvement for the Opus tier, powering long-running agents with better coding and professional work.
Source: Anthropic News（官网监测） · 2026-07-25 · Original
2. Quoting Boris Cherny
Boris Cherny says Opus 5 is the least prompt injectable model yet, based on PI evals and red teaming.
Source: Simon Willison's Weblog · 2026-07-25 · Original
3. Introducing Claude Opus 5
Claude Opus 5 offers frontier intelligence close to Fable 5 at half the price, leading the Artificial Analysis leaderboard and showing autonomous proactivity.
Source: Simon Willison's Weblog · 2026-07-24 · Original
4. Supporting ambitious external research through the Anthropic Economic Futures Research Fund
Anthropic commits $200 million to the Economic Futures Research Fund to support ambitious external research.
Source: Anthropic News（官网监测） · 2026-07-24 · Original
5. Inside the Model Factory — Eiso Kant, Poolside AI
Poolside AI's Laguna S 2.1 outperforms models nearly 10x its size, and its co-founder details the Model Factory that releases models in eight weeks.
Source: Latent Space · 2026-07-23 · 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._