Story perspectives

TypeSafe AI launches Jev, gets $40M funding

By Drooid · Sep 18, 2026, 10:09 PM UTC · How we work

Summary

TypeSafe AI Launches Jev

TypeSafe AI launched Jev, a non-text model that returns calibrated decisions.
Jev processes inputs in 70-500 ms, costing $0.042 per million tokens, 10-20× cheaper than Gemini and Anthropic’s Fable 5.1.
Vercel replaced OpenAI’s Luna 5.6 with Jev, achieving 5-18× faster response and higher accuracy.
TypeSafe AI announced $40 million funding alongside Jev’s launch.
Almeida predicts cheap intelligence will spread like the early internet, embedding smart software everywhere.

Summary

AppsFlyer Adds ChatGPT Attribution

AppsFlyer added ChatGPT Ads attribution via OpenAI’s Conversions API, letting marketers measure AI traffic.
The integration provides deduplicated reporting with no engineering needed.
Attribution now includes eight AI answer engines and 13 search engines free of charge.
OpenAI’s ChatGPT Ads app lets U.S. Shopify merchants sync stores, set budgets, and run ads in the chatbot.
Meta’s Muse AI agent app displaced ChatGPT to top the U.S. iPhone App Store chart.

Summary

Spirit AI Deploys Robots

Spirit AI plans industrial robot deployment within two years, citing 90% success on simple tasks.
Co-founder Gao Yang says the robot brain, not hardware, remains the stack’s weakest link.
Around 1,000 contractors wear sensors at homes and factories to gather motion data for training.
Robots use whole-body force control and automatic emergency braking to improve safety in crowded spaces.

History

Historical Context

Amazon Web Services launched S3 on March 14 2006 and EC2 in August 2006, slashing compute costs to $0.12 per hour and attracting early adopters such as Netflix. Its pay-as-you-go model grew to a 31% cloud-infrastructure share by 2023, enabling startups like TypeSafe AI and AppsFlyer to offer AI decisions at $0.042 per million tokens, accelerating affordable AI across industries.

1 of 4

TypeSafe AI Launches Jev

TypeSafe AI launched Jev, a non-text model that returns calibrated decisions.
Jev processes inputs in 70-500 ms, costing $0.042 per million tokens, 10-20× cheaper than Gemini and Anthropic’s Fable 5.1.
Vercel replaced OpenAI’s Luna 5.6 with Jev, achieving 5-18× faster response and higher accuracy.
TypeSafe AI announced $40 million funding alongside Jev’s launch.
Almeida predicts cheap intelligence will spread like the early internet, embedding smart software everywhere.

1 / 4