CommunityJev Media & Talks
Thirteen questions, one document
AIEdTalks part 4 on the batching trick: a loop that resends a long document for every question, versus one call that asks all thirteen.
- Format
- Video
- Published by
- Community
- Author
- AIEdTalks
- Last verified
- Oct 2026
Works forJev for Classification
Used inJev for Software & DevTools
Highlights
- Framed as part 4 of a series on Jev, specifically the batching trick.
- The description contrasts a per-question loop, which resends the document, with one request that holds every question.
- The thirteen-question shape matches TypeSafe's parallel-questions cookbook.
Watch out
A series episode. The measured 12.2x cost and 10.0x speed figures are in the cookbook, not restated here.
Questions about this entry
Short answers drawn from the linked project or page.
The repeated document upload. A per-question loop resends the whole document each time, while one request holding every question sends it once. TypeSafe's docs say questions are evaluated in parallel against the same state, so extras barely change response time.
They are not in the video. The entry points to TypeSafe's parallel-questions cookbook as the source of those measured figures, and this episode restates the shape rather than the measurements, so check the cookbook for method and payload.
Parallel isolated evaluation is what makes it cheap, but the trade-offs are real: failures get harder to attribute and one oversized payload can hit request limits. The episode does not discuss either cost.
It is part four of a series and assumes the earlier setup. The batching pattern stands alone, but the video does not document authentication, model selection, or the basic request shape, so keep the cookbook alongside it.
It is described as part four of a series on Jev, focused specifically on the batching trick. Because it assumes the earlier setup, it does not cover authentication, model selection or the basic request shape.
The thirteen-questions-in-one-call shape matches TypeSafe's parallel-questions cookbook. The measured 12.2x cost and 10.0x speed figures live in that cookbook, not in the episode, so check it for method and payload.
The loop resends the whole long document for every question, while one request holding all thirteen sends it once. TypeSafe's docs say questions are evaluated in parallel against the same state, so extras barely change response time.
Reactions & coverage
Posts, threads, and videos about this entry from around the web.
YouTube: AIEdTalks on asking thirteen questions in one call instead of resending the document each time.
More Jev resources like this
Max Jesch, on IBM Developer, walks Jev as a System One model that returns probabilities for classification instead of generating text.
Jev Media & TalksVideo#community#video#explainer
Community
Martin Keen, on IBM Technology, explains Jev as a System One model whose output is a decision rather than generated text.
Jev Media & TalksVideo#community#video#explainer
Community
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
Fireship on Diogo Almeida's System One launch claims: no chat output, and the headline 200× speed and 400× cost multiples.
Jev Media & TalksVideo#community#video#explainer
Community
Bloomberg News clip on TypeSafe's non-chat model and its promise of a cheaper, faster alternative to large language models.
Jev Media & TalksVideo#community#video#media
Community
Patrick Desjardins wires Jev into Cursor for deterministic decisions from the public docs, with blog posts on hundreds of text rules under two seconds.
Jev Media & TalksVideo#community#video#tutorial
Community
From the Jev community
Posts from builders shipping with Jev right now.