Liquid AI's d1-ultra blocks only dangerous commands in coding agent, no recalibration needed
JosephJacks_ · x · 2026-10-17
A developer running an auto mode plugin for OpenCode (similar to Claude Code's Auto Mode) found that Jev 1.14 kept blocking read-only commands, forcing repeated safety recalibration. After switching to Liquid AI's newly released decision model d1-ultra, with zero recalibration it blocked only the genuinely dangerous command. d1-ultra is Liquid AI's most powerful multimodal decision model, supports 44 languages, ranks #1 on the Artificial Analysis Decision Index (65), and leads all eight categories of its 112-evaluation suite.
More from coding & agent
- iOS Devs: Install These Agent Skills to Make Your App Designs Stand Out — stefanjblos · 2026-10-17
- Devin Can Now Build and Test iPhone Duo Apps in Cloud VMs with Video Verification — DevinAI · 2026-10-17
- AutoSep cuts agent reward hacking from 69.9% to 12.1% via weak-to-strong verifier hardening — Stanford · 2026-10-17
- Free open-source course teaches harness engineering with 14 lectures and 8 hands-on agent projects — techNmak · 2026-10-17
- Anthropic engineer: Claude Slides is now how I make slide decks — bcherny · 2026-10-17
- Dev built soft-body physics engine with Claude's help, claims it tops BeamNG — grandprixlegends · 2026-10-17