Dropstone 1.5 launches terminal coding agent

Re-baselining runtimes based on a monthly shootout of open-weight models is a brilliant arbitrage play on LLM price-performance drops, but its ultimate viability depends on how seamlessly the agent handles prompt-engineering differences between models.

* Monthly model swapping ensures users always have access to the absolute best open-weight frontier model without manual migrations.

* Pricing of $15/mo for 450 sessions/week is extremely aggressive and undercuts proprietary giants like Claude Code Pro by offering twice the volume at a lower price.

* Agent prompt optimization is highly model-dependent; generic agent prompts might degrade performance compared to vertically-integrated solutions.

* US-hosted infrastructure and zero-retention policy address enterprise data governance concerns.

* Explicit user confirmation for terminal commands and file writes provides a vital safety guardrail for agentic execution.