OpenRouter's Auto Code router picks the model for each coding call
It weighs price, provider health and benchmarks per request — as Redis ships Iris for agent memory and DigitalOcean brings agent oversight to your phone.

Copy markdown
Auto Code picks your coding model per request
OpenRouter's new Auto Code router chooses the model and reasoning effort for each request, weighing eleven live signals — current pricing, provider health, benchmarks and cache state — and only from models your governance policy allows. Point a coding agent at it and you stop hand-picking models while it optimizes cost against quality call by call.
Redis launches Iris, an agent context engine
Redis reframed its in-memory store as a context-and-memory platform for agents, shipping six parts: Intent Router, Agent Playbooks, Agent Memory, Context Gateway, Lens and Agent Mesh — routing, reusable procedures, persistent memory, observability and agent-to-agent messaging. If you already run Redis, that's a memory and context layer you no longer have to assemble from scratch.
Supervise your agents from an iPhone
DigitalOcean shipped Cuttlefish, an iOS app for its Managed Agents: watch live agent sessions, approve or deny individual tool calls, and stop a run remotely. Managed Agents is still in public preview, but you can now babysit a long-running agent from your phone instead of a terminal.