Four major AI models landed in three days: Tencent's Hunyuan Hy3, xAI's Grok 4.5, Meta's Muse Spark 1.1 and OpenAI's GPT-5.6. Anthropic returned from a short-lived export-control suspension. And there is no single winner any more: leadership on the frontier now depends on the workload and the evaluation.

This episode includes the complete practical guide, in full: which AI to use for which task across six workloads (software engineering, terminal and infrastructure, high-volume automation, computer-use agents, science and healthcare, on-premises deployment), plus the four-step evaluation to run before routing production traffic.

Also inside: my Yorkshire Post column on walking Boston's Freedom Trail with an AI Harvard historian in my ear; the launch of The Signal Playbook, a free series of production-ready prompt guides; and the Week 27 layoff tracker, where the mid-year numbers are stark: 585,151+ cross-sector job losses in 2026, 185,894 in tech, Microsoft cutting 6,400 including 20% of Xbox, and Oracle's 25,254 locked in via the 10-K.

The full written edition, with every comparison table and the tracker graphic, is free at https://thesundaysignal.ai

David Richards MBE is a technology entrepreneur, co-founder of Yorkshire AI Labs, and weekly columnist for the Yorkshire Post. Episode tags: AI, artificial intelligence, tech layoffs, Claude, OpenAI, GPT-5.6, Grok, Meta, Tencent, agentic AI, enterprise AI, future of work, tech news, Microsoft, Oracle, Xbox

Season/Episode: Issue 62

Publication date: Sunday 12 July 2026