Summary
Matt Wolfe breaks down two major AI model launches that arrived within hours of each other: Anthropic’s Claude Opus 5.5 and OpenAI’s GPT-6-Sol and GPT-6-Luna. He walks through Anthropic’s claims that Opus 5.5 matches Claude Fable 5.1 on most tasks while costing 40% less to run than Opus 5, then dives into benchmark results across Terminal Bench 4.0, Frontier Code, Cursor Bench, GDPval, and Artificial Analysis’s intelligence index, where Opus 5.5 posts a new top score of 58.
Wolfe also examines real pricing shifts — Opus 5.5 dropping to $4 per million input tokens and $20 per million output tokens, down from Opus 5’s $5/$25 — and explains why per-task costs haven’t fallen as sharply due to higher token usage per task. He showcases community-built demos using Opus 5.5, including JavaScript-coded animations, a Dark Souls-style game clone from Forward Future, and a flight simulator, illustrating how far agentic coding and generative game development have progressed.
The video then turns to OpenAI’s response: GPT-6-Sol and GPT-6-Luna, launched roughly two hours after Opus 5.5. Wolfe compares their pricing cuts — GPT-6-Sol’s input cost halved to $2 per million tokens — against benchmark results showing the new OpenAI models trailing Opus 5.5 but undercutting it on price, giving viewers a clear snapshot of the current competitive landscape among Anthropic, OpenAI, and their evolving cost-performance tradeoffs.
📺 Source: Matt Wolfe · Published September 22, 2026
🏷️ Format: News Analysis
