Dev trains his own Jev-style API with one prompt for ~$6.60 in 16 minutes on an A100
Gradio · x · 2026-10-05
After OpenAI, Liquid, and Nace all shipped Jev-style APIs this week, developer MaziyarPanahi trained his own equivalent with a single prompt in HuggingChat:
- An agent ("ML Intern") ran the training on an NVIDIA A100 in 16 minutes
- The agent fixed its own bugs along the way
- The full 60K version cost about $6.60 and picked the right tool 62% of the time (20 tasks)
A neat demonstration of agent-driven small-model training, which the official Gradio account reposted.
More from coding & agent
- Comparing 4 AI Animation Pipelines with Opus 5.5: Token Usage Turns Out Similar — chongdashu · 2026-10-06
- Build game worlds by walking around and leaving notes, in real time with Spawn — TAbrodi · 2026-10-06
- Teknium catalogs 305 open, hackable devices your AI agent can control and live in — Teknium · 2026-10-06
- Crawler + GPT agent scanned 5,137 jobs for $6.53, found 3 low-competition niches — Arindam_1729 · 2026-10-06
- xAI Cookbook adds 4 Grok SDK examples: screenshot-to-React, X sentiment tracker, AI ad generator — tetsuoai · 2026-10-06
- Power user runs a swarm of personal agents — poke, instinct, muse, dot, Grok bot, pickle and more — sebkrier · 2026-10-06