Are the speed and cost figures measured or reported?

Reported. The entry attributes them to TypeSafe results on selected workflows, and the description says the animation speed in these demos is not a model-processing benchmark. Nothing on the page is an independent measurement of latency or spend.

Why do the Wikipedia race numbers look better than the real site?

Because page loading is excluded from the timings, per the description. That is a reasonable choice for isolating the decision call and a misleading one for estimating end-to-end wall clock, so do not carry those figures into a latency budget.

The candy demo says 150,000, the narration says 100,000. Which?

The page does not reconcile the two numbers, and the highlights flag the mismatch explicitly. Treat the item count as unverified and the throughput implied by it as unreliable.

Is there anything I can actually copy from this?

A written quick-start at graysonhyc/ai-guides. The demos themselves are shown on video with no repository linked for them, and the channel covers five separate demos, so this is a tour rather than a tutorial.

Which demos does the walkthrough cover?

Colour sorting, an AI town, a Wikipedia race, model routing, and games. The entry tags the clip for routing and control work across software and gaming.

Who made the video, and what does it caution?

Grayson Ho published it. The entry notes the description says the animation speed in these demos is not a model-processing benchmark.

What written resource accompanies the demos?

A written quick-start at graysonhyc/ai-guides. The demos themselves are shown on video, and the entry records no repository linked for them.

CommunityJev Media & Talks

Decisions inside apps

Grayson Ho walks colour sorting, an AI town, a Wikipedia race, model routing, and games, and separates narration from the measured parts.

See highlights

Open youtube.com

Format: Video
Published by: Community
Author: Grayson Ho
Last verified: Sep 2026

Tagscommunity video demos

Works forJev for Routing Jev for Realtime control

Used inJev for Software & DevTools Jev for Gaming

Highlights

Covers colour sorting, an AI town, a Wikipedia race, model routing, and games.
Speed and cost lines are TypeSafe-reported results on selected workflows.
Wikipedia timings exclude page loading. The sorting demo shows 150,000 virtual sweets; the narration says 100,000.
A written quick-start lives at graysonhyc/ai-guides.

Watch out

The description says the animation speed is not a model-processing benchmark.

Questions about this entry

Short answers drawn from the linked project or page.

Reactions & coverage

Posts, threads, and videos about this entry from around the web.

YouTube: Grayson Ho walks sorting, a Wikipedia race, and model routing, and flags which numbers are vendor-reported.

More Jev resources like this

What people built in 72 hours

AI with Rithesh rounds up early Jev loops: a drone, games, a browser, email sorting, and one labelled product benchmark.

Jev Media & TalksVideo#community#video#demos

Community

System 1 and System 2, for models

LangChain maps Kahneman's split onto models: System 1 as the fast automatic decision, System 2 as the slow deliberate one, with Jev in the first role.

Jev Media & TalksVideo#community#video#explainer

Community

A September walkthrough with live demos

AiNow dates the Jev release to September 2026 and walks live demos of a model that returns a decision instead of the next token.

Jev Media & TalksVideo#community#video#demos

Community

Where a decision model sits in the architecture

Aishwarya Srinivasan's 22-minute end-to-end on using System One models inside software, instead of sending every structured check to a text model.

Jev Media & TalksVideo#community#video#architecture

Community

a16z with Diogo Almeida on software that decides

Ben Horowitz and Martin Casado interview Diogo Almeida on why capable models have not automated ordinary software, and what Jev is for instead.