Anthropic has released Claude Opus 4.7, achieving an 80.1 score on the SWE-Bench Verified benchmark, a minor decrease from its predecessor. This latest version emphasizes safety tuning, potentially at the expense of peak benchmark performance. Separately, a developer has created a persistent coding workspace for Claude AI costing $10 per month, utilizing Pi's execution layer and Cloudflare Tunnel to overcome context limitations. AI
影响 Anthropic's latest model prioritizes safety, potentially impacting its competitive edge on benchmarks, while a new low-cost workspace emerges for Claude AI users.
排序理由 The cluster contains a new model release with benchmark scores and a technical implementation for a persistent workspace.
在 Mastodon — fosstodon.org 阅读 →
AI 生成摘要 · Google Gemini · 来自 3 个来源。 我们如何撰写摘要 →