x.com
paulwei on X: 昨天出的 Jev 模型
@coolish@coolishlisted 51m agoreviewed by Jev昨天出的 https://typesafe.ai Jev 模型 适合高频决策，那岂不又是适合打游戏？ 之前用 GPT-6 Astra 代打杀戮尖塔2 能力强但速度慢。 我刚实测用 Jev 打，行动思考只需 0.7秒， 画面我都没看清它就操作完了。 这超人类游戏速度， 着实又让我震惊瘫坐了😅 语言很难描述这感觉，看视频⬇️ ❤️ 1.1K likes on X
- Author
- @coolish
- Use case
- Games & Real Time
- Added
- 2026-09-25
All figures come from the author. Check the source before you quote them.
More in Games & Real Time
- GitHub - KyleKreuter/jev2048: Let Jev (TypeSafeAI) solve 2048▲ 1github.com
- Browser Use on X: Luna (planner) -> Jev (actor) playing poker and winning▲ 0x.com
- Thinh Le on X: Can someone make Laya and Jev work? 😅▲ 0x.com
- Arthur Marques on X: @Gautamk104 @typesafeai was trying to figure out how to make these 2 models figh▲ 0x.com
- Geek Lite on X: 把本地 MLX 推理的 Laya 模型和托管 API 的 Jev 放在同一个 T-Rex 跑酷赛道上对打，用延迟、存活和回放数据对比本地推理与云 API。▲ 0x.com
- Koimiao🐈 on X: We built a Stanford Town powered by Jev.▲ 0x.com
Jev guides for this use case
- Jev in the agent loop: routing, guardrails, contextRouting, tool-call guardrails and context compaction — the three places a decision model earns its keep inside an agent.
- Jev vs LLM: a decision model is not a smaller text modelWhy a decision model and a text model aren't substitutes, where the boundary sits, and the cost math.
- Is Jev reliable? The honest limitationsWhat the no-hallucination claim really covers, where the accuracy ceiling sits, and the failure mode we hit ourselves.
Bid history
No bids yet — the first one takes this project straight to the spotlight.