← All issues

    AgentVet Weekly — July 13, 2026

    July 13, 2026

    From the founder

    The story this week isn't really about GPT-5.6's three-tier naming scheme or Claude going mobile — it's about infrastructure catching up to ambition. When a cloud provider's CTO is reframing their entire product around 'Agent Experience,' and OpenAI is tiering models by cost and capability specifically to fit agent pipelines, the underlying message is the same: the people building the picks and shovels have decided agentic workloads are the permanent workload, not a niche. That shift in supplier conviction is worth more attention than any individual benchmark.

    Simon Willison

    The new GPT-5.6 family: Luna, Terra, Sol

    OpenAI's latest flagship model hit general availability this morning , and comes in three sizes: Luna, Terra, and Sol (from smallest to largest). The new models are priced per 1M input/output tokens as Luna $1/$6, Terra $2.

    Latent Space

    Why AI Infrastructure must evolve for Agent Experience — Akshat Bubna, Modal CTO

    Modal CTO Akshat Bubna argues that cloud infrastructure needs a ground-up rethink to serve agent workloads — specifically around latency, stateful execution, and the cost profile of many short bursty calls versus long batch jobs. Two years of building what he calls an 'agent cloud' gives this more weight than typical vendor positioning; the lessons are concrete and relevant to anyone choosing or building execution infrastructure for agents. If you're evaluating where to run agents in production, this is required listening.

    Get AgentVet Weekly in your inbox

    The latest in agentic & applied AI, every week.

    ← All issues