Descriptions:
Alex Finn runs an extended live stream centered on evaluating Claude Opus 4.7, opening with a pointed defense of Anthropic’s models against widespread negative sentiment on X. Finn argues that Anthropic is the AI company least likely to have influenced his coverage — receiving no early access, free subscriptions, or outreach compared to OpenAI and Google — and frames the session as an unbiased benchmark test to determine whether critics are right or simply reacting to company allegiances.
The core of the stream is the “world famous Alex Finn benchmark,” a custom coding test applied to Opus 4.7, alongside a live review of Claude Design for UI generation and a first look at OpenAI’s newly released Codex desktop application for computer-use automation. Finn also covers Perplexity’s personal computer feature. The stream captures real-time troubleshooting when Claude Code stops responding — which Finn attributes to a platform outage — generating candid commentary about infrastructure reliability and whether OpenAI’s aggressive GPU investment gives it a long-term stability advantage.
For viewers following the mid-2026 model competition, this episode offers a practitioner’s real-time perspective on Opus 4.7’s coding performance, the UX of Claude Design, and the emerging category of desktop AI agents with the Codex super app release.
📺 Source: Alex Finn · Published April 17, 2026
🏷️ Format: Livestream
