Termexo v0.8.2: smoother terminal input and new model controls
Maker update: Termexo v0.8.2 is out today. This release focuses on typing responsiveness while multiple agents produce output.
Output now uses one application-wide subscription with routing by terminal ID, reducing duplicate callbacks and UI work. History replay restores the terminal display without reprocessing old output through task, handoff, and startup readers. Draft persistence is batched, and terminal size ownership is handled on focus/click instead of every keystroke.
Also new: Claude 1M context configuration, separate Claude/Codex effort settings, per-terminal launch overrides, and direct agent buttons in empty workspaces. Actual context and effort support depends on the CLI, model, account, and provider.
Termexo is MIT-licensed and runs on Windows. Agent CLIs and model services need their own setup. Release notes and installers: https://github.com/gemron/Termexo/releases/tag/v0.8.2
English guide: https://www.termexo.com/guide.en.html
AI-written from the official release notes. No benchmark multiplier is claimed.
Replies
i like the idea of separate effort settings for claude and codex. different tasks usually need different levels of reasoning.
Termexo
@chandrshekhar_rawen Exactly—that's why the settings are separate. You can keep a preferred effort setting for each CLI, then override it when launching a particular terminal. For example, a small fix and a more involved investigation don't have to start with the same setting. The available effort levels still depend on the selected CLI and model.
@guomengyue1987_beep Thanks for clarifying! That flexibility makes a lot of sense being able to set defaults per CLI and override them on the fly for quick fixes versus deep dives will definitely save time and compute.
Per terminal launch overrides could be handy for people running different agents across multiple workspaces.
Termexo
@uttam_kumar35Â Yes, that's the use case. The launch dialog lets you choose settings for that terminal while leaving the saved profile defaults in place. You can use a different effort level for a one-off task without changing the default for later launches. Leaving an override unset uses the profile's setting, which helps when several terminals need different setups.
@guomengyue1987_beep Seamlessly switching between quick edits and complex agent runs without touching global settings is a huge UX win.
The focus change for terminal size ownership instead of every keystroke is clever. I noticed similar lag before. Does history replay feel instant now. Maybe add a short comparison video to show it.
Termexo
@devinstone Thanks, Devin! I wouldn't call it instant without measuring it on the same workload. In v0.8.2, replay stopped feeding old output back through the task and startup readers. Since then, v0.8.6 moved reconnection to screen-state restoration, and v0.8.7 fixed blank-screen and redraw-blocking issues. A side-by-side recording with the same history and terminal count would be a useful way to show the difference—thanks for the suggestion.
The 1M context option is interesting. i would mainly want to know how much of that is actually usable through different providers.
Termexo
@thomas_jack3Â That's the key limitation: the 1M option is a Claude launch configuration, not a guarantee of a 1M window from every provider. The usable context depends on the model, CLI version, account access and provider support; Termexo can't raise those limits. I don't have a verified provider-by-provider compatibility table to share. Which provider and model are you considering? That would let us check a specific combination rather than assume the toggle means full support.
@guomengyue1987_beep Fair point I was mostly curious about using it with Anthropic's direct API vs. OpenRouter. Good to know it relies on what the underlying provider actually supports.