A week that balanced app-layer momentum with two major model releases.
Perplexity's Comet expands
On February 20, Perplexity's changelog opened Comet browser pre-orders for iOS and upgraded the Comet Agent — now defaulting to Claude Sonnet 4.6 (Pro) / Opus 4.6 (Max) — with enterprise Memory, response customization, and a Finance module adding analyst ratings and SEC-filing deep links. The research agent keeps turning into a full workspace.
Ambience scales across a hospital system
On February 17, Ambience Healthcare went enterprise-wide at Houston Methodist, spanning ambulatory, emergency, and inpatient settings — a marker of clinical AI moving from pilots to system-wide deployment.
Model watch: Claude Sonnet 4.6 and Gemini 3.1 Pro
The base models both jumped. On February 17, Anthropic shipped Claude Sonnet 4.6 (a full upgrade across coding, computer use, and agentic planning; 1M-token context beta; held at $3/$15). Two days later (February 19), Google launched Gemini 3.1 Pro, citing a 2×+ reasoning gain at no price increase. Every agent built on them inherits the gains.
What it means if you're picking tools
Research agents, clinical AI, and the models underneath everything all advanced. See research, chat assistants, coding, and transcription.
Sources: Perplexity changelog (Feb 20) · Ambience × Houston Methodist · Claude Sonnet 4.6 (Anthropic) · Gemini 3.1 Pro (Google)