signalLenny's Newsletter2026-09-29
🎙️ How I AI: Jev for beginners + I left Claude for months, Opus 5.5 brought me back + Opus 5.5 vs. GPT-6 Sol bench
Claire tests Jev, TypeSafe AI's decision model that returns structured choices, scores, and probabilities instead of generated text. She analyzes 1,700 pull requests for 9 cents, processes 200,000 classifications for about $4, and finds Jev covers roughly 90% of software workflow needs at 4 cents per million input tokens with no output-token fee. Jev works best alongside a frontier model, and its speed makes real-time AI loops practical.
- for who
- Developers and AI practitioners building large-scale classification or decision pipelines
- why now
- Jev's new model at 4 cents per million tokens makes large-scale real-time apps practical now.
- what changes
- They can shift expensive language-model tasks to a cheap decision model, making real-time and bulk analysis workflows feasible
- to do
- Identify workflow steps that only need a decision, then route them to Jev while sending only important groups to a frontier model
key points
- Jev returns predefined values, not text, at 4 cents per million input tokens with no output fee
- Claire analyzed 1,700 pull requests for 9 cents and 200,000 classifications for about $4
- Jev complements frontier models: classify and route with Jev, deep reason with GPT-6 Astra
#jev#decision model#cost optimization#workflow
score
score 9 out of 10. 0-10: how dense the facts are, multiplied by how much you can do with them after reading. 8+ means the topic's evidence bar is met: benchmarks and availability for a new model, amount and investors for a funding round, revenue figures for a solo-money story. Below 5 an item does not enter the digest. A press release scores 3 or less, a reprint loses 2, anything older than 14 days loses 1, a headline that misleads loses 3.
read the source