Computer use built on Jev
X post by Aaron Levin describing a computer-use agent built on Jev, with claimed cost and speed advantages over a frontier LLM.
This summary is based on how curated lists describe the project; its README was not available to review.
Overview
This is a social media post in which Aaron Levin shares a computer-use setup that uses Jev as the decision maker. According to the list summary, he reports it as about 155x cheaper than Opus 5, roughly 20x faster, and generalising across operating systems.
The post is useful as an informal data point on using a decision model to choose on-screen actions instead of having a large LLM write tool calls. No code or method is available in the directory record, so the figures are the author's own claims and have not been independently checked.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Choice | Which on-screen action or element should the agent use next?(inferred) | — |
How it works
- Capture the current screen or UI state
- Ask Jev to choose the next action from allowed options
- Execute the action and repeat
Strengths
- Author reports large cost and speed gains versus a frontier LLM
- Claimed to work across operating systems
Caveats
- List-only; no code or benchmark details available
- Figures are the author's claims
At a glance
- Type
- Discussion
- Status
- discussion
Use cases and integrations
How curated lists describe it
Aaron Levin: 155x cheaper than Opus 5, about 20x faster, and it generalizes across operating systems.
hellogumbo/awesome-jev
Frequently asked questions
What does the post claim?
That a Jev-based computer-use agent is about 155x cheaper than Opus 5 and around 20x faster, and works across operating systems.
Is code available?
The directory record links only the post; no repository is listed.
How does Jev fit?
It would choose actions from typed options rather than generating tool calls as text.
Related projects
Auto-updated bilingual leaderboard of Jev ecosystem projects, tutorials and discussions, discovered daily and sorted by GitHub stars.
ProjectproductionChatGPT-style web chat where every word of every reply is a Jev multiple-choice decision over a ~1,800-word vocabulary plus END.
ProjectprototypeTypeScriptExperiment that deliberately misuses Jev as a text generator by asking which symbol comes next and sampling from the returned distribution.
ProjectresearchPythonChat client that makes Jev write by asking Next word? over a 20,000-word dictionary in 255-word buckets plus a run-off.
ProjectresearchPythonIssue in the Orcha multi-agent orchestration project proposing to move judgment calls onto Jev to cut model cost.
ProjectdiscussionPythonAuto-generated write-up of a Reddit post claiming an open RLCD-trained Jev model; content is generic and largely inaccurate.
ProjectPython