chat: stop, regenerate, inline prompt editing, copy answer

images on web-search answers, via Tavily

web search: tell verifier answer is web-grounded; add 'searching' stage

tier-1 lineup: Gemma 4 (general), Qwen3 Coder (coding), DeepSeek V4 Flash (logic/math)

tier-1: DeepSeek V4 Flash (general), Qwen (coding), Skyfall 36B V2 (creative)

deploy hardening: access gate, rate limiting, spend guard, env CORS

hard gate: GPT-5 reserved exclusively for hard STEM/reasoning queries

stream the hedge draft live from t=0; slow the typewriter to reading pace

hedged speculative draft with user topic toggle

progressive boss-fight status: reasoning ticker with arcade fallback phrases

checkpoint: current behavior before latency work

AuctionRouter: cost-aware multi-agent LLM router with auction-based selection and verification-gated escalation