Jev API calls seem prone to failing
The user notes that Jev's API calls appear prone to failures, speculating that low API costs lead to heavy usage and server overload.
THE JEV GUIDE · CURATED EDITION
Launch news, explanations, demos, and community projects around TypeSafe Jev, organized by topic. Every entry links back to its X source.
Curated selection · Not exhaustive · Updated manually
Sponsors
These products pay for the directory. Want yours up here?
The user notes that Jev's API calls appear prone to failures, speculating that low API costs lead to heavy usage and server overload.
Ollama now supports TypeSafe's Jev decision model API with a /v1/systemone endpoint, enabling yes/no, multiple choice, or scored responses, fully local with no network cost. Nimble 9B runs at 91ms per decision on M5 Max, with three models available.
TypeSafe AI reportedly closed a $40M seed round led by DCVC at ~$200M on Sept 15; nine days later it was in talks at a $10B+ valuation. Within a month of launch, 1% of businesses on Ramp were paying for its Jev model. Link
Larry Gadea hints that more Jev credits may be needed, suggesting growing usage or rapid consumption.
Langfuse JP community announces support for Jev and OpenAI decision models, enabling high-speed and low-cost judgments.
One card in the feed, every sixteen, on every page of it. $50 a week, cancel any week.
According to the post, Jev is used to judge source relevance. In a 28-task comparison, relevant information increased while costs remained controlled, reflecting an improvement in post-retrieval filtering.
Someone predicts OpenAI Decisions API might be an instant hit thanks to image support, posing a threat to Jev.
Marco De Rossi welcomes OpenAI to the board, sharing a link to the Jev models API and tagging #DecisionModels and @LevantoLabs. Link
Jev is not a chatbot but a decision AI that provides probabilities. It has now been integrated into Alarm Monster, allowing users to simply type /jev in the bot without writing API code. In a test example, Jev assessed the chance of NMR replicating QNT's 5x move at 40%, with a cost of about $0.00004 per answer.
Citing a WSJ report, the post notes TypeSafe AI's Jev decision model is used by 25% of Fortune 500 companies and has sparked imitators, while pointing to alternatives from Cloudflare, AWS, and Databricks.
Replit announces new features to create beautiful data visuals, build with GPT-6.1 and Claude Sonnet 5.5, and add Jev to your apps without requiring API keys.
NVIDIA and Arcee explained at AI Engineer World's Fair why companies moved to open models, noting that Jev signups are paused while open copies like Kev already run on a laptop GPU.
Star the project on GitHub to support continued updates and curation.