# A weekend with Jev made my coding agents up to 31% faster

**URL:** <https://forum.devtalk.com/t/a-weekend-with-jev-made-my-coding-agents-up-to-31-faster/249954>\
**Category:** AI In The News\
**Tags:** via-news-bot, coding, jev\
**Created:** [21 September 2026 15:47 UTC](https://forum.devtalk.com/t/a-weekend-with-jev-made-my-coding-agents-up-to-31-faster/249954 "2026-09-21T15:47:04Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![CommunityNews](https://forum.devtalk.com/user_avatar/forum.devtalk.com/communitynews/32/1113_2.png) [@CommunityNews](https://forum.devtalk.com/u/CommunityNews)\
**Post date:** [21 September 2026 15:47 UTC](https://forum.devtalk.com/t/a-weekend-with-jev-made-my-coding-agents-up-to-31-faster/249954/1 "2026-09-21T15:47:04Z")

</div>

> I spent a weekend playing with Jev, a model that scores things instead of writing text. It became a code search tool that makes Codex, OpenCode and Claude Code faster, measured over 162 sessions.

Read in full here:

> **[A weekend with Jev made my coding agents up to 31% faster](https://tyrpien.com/blog/oko-agent-search)**
>
> I spent a weekend playing with Jev, a model that scores things instead of writing text. It became a code search tool that makes Codex, OpenCode and Claude Code faster, measured over 162 sessions.

---

<div class="post-metadata">

**Author:** ![mira\_tokarev](https://forum.devtalk.com/letter_avatar_proxy/v4/letter/m/d26b3c/32.png) [@mira\_tokarev](https://forum.devtalk.com/u/mira_tokarev)\
**Post date:** [24 September 2026 03:50 UTC](https://forum.devtalk.com/t/a-weekend-with-jev-made-my-coding-agents-up-to-31-faster/249954/2 "2026-09-24T03:50:02Z")

</div>

The input-cost angle seems especially interesting for a decision model, because repeated tool observations can dominate even when the model emits little or no text. Before changing the scorer, I’d try keeping the same search policy and applying a deterministic local compression pass to the serialized agent state, then measure decision agreement, search quality, and input tokens per decision. Contextpress ([GitHub - Taha-azizi/contextpress: Deterministic context compression for LLM chat, RAG, and agent pipelines · GitHub](https://github.com/Taha-azizi/contextpress) · [contextpress · PyPI](https://pypi.org/project/contextpress/)) is a small way to test that boundary without claiming that compression is universally safe. If the decision model’s probabilities remain calibrated while duplicate observations fall, that would be a useful complement to the 162-session speed result.