vibehacker
News
Cursor Docs ·

DeepSeek V4.1 Flash lands in Cursor's model picker at $0.60 to $0.98 per CursorBench task

Added Oct 26 with no BYOK proxy needed, the open-weight 1M-context model turns on under Cursor Settings > Models, and Cursor says all four effort tiers (44.9% to 47.7% on CursorBench 4.0) sit on its cost frontier. It still trails Opus 5.5 Max's 57.8%, and as a third-party model it bills outside your plan's included usage.

More news

View all

Bito's Governor routes Claude Code, Cursor and Codex requests by task risk and feeds them code context

Launched Oct 27, you point your agent's base URL at Governor, which picks a model and reasoning level per request from the models your team allows, holds that choice for the whole task, and answers code searches itself with file and line results. Bito reports one customer's Claude Code cost per task fell 48% ($4.12 to $2.14) with success held at 100%; access is by request…

Bito Blog

Factory Droid 0.250.0 shows each session's PR state and asks before its browser panel opens other apps

Released Oct 27 with Desktop 0.207.0, sessions now show whether their pull request is a draft, open, merged or closed, and long sent messages collapse behind Show more. The browser panel now denies device access and asks before opening another app, opening a saved session from another folder asks for folder trust first, and Slack only opens a channel for real code changes…

Factory Changelog

Copilot CLI 1.0.104 lets Autopilot wait on background tasks and drops the experimental flag for Assisted Permissions

Released Oct 27, interactive and SDK Autopilot sessions now keep waiting for attached background tasks instead of quitting at COPILOT TASK WAIT TIMEOUT SECONDS ( p autopilot still honors it), and Assisted Permissions work without turning on experimental mode. Auto also falls back to its default when a saved preference disappears, instead of failing every turn…

Claude Code 2.1.311 compacts past 200K tokens instead of stopping, and lets ultrathink raise effort mid-turn

Released Oct 26, standard context sessions that run over 200K tokens now compact and keep going instead of stopping with "Usage credits required for 1M context", and typing ultrathink while Claude works bumps the rest of that turn to xhigh effort. It also fixes startup freezes from long CLAUDE.md list items, Write and Edit stalling on huge files, and subagents missing MCP tools from servers that connect late…

Spotted something we missed? Start a thread.