Fantail SLMs for Coding Agents

Show HN: Fantail SLMs for Coding Agents

Operating in show hn launch, Fantail SLMs for Coding Agents operates.

Profile

We’re releasing Fantail, small language models tuned for coding agents with schema‑first prompting and constrained decoding. On Terminal‑Bench: 0.5B 28.4%, 1.3B 34.7%, 3B 41.3% vs GPT‑5 mini 39.8% and Sonnet 4.5 45.7% (3 seeds, no web). Reproducible runs + CSV→figure script included.PS: we will release the weights and details later this week.LinksPaper/overview: We will be posting here the details later this week. Weights: https://huggingface.co/autohandai/models Code: https://github.com/autohandai/fantail

Fantail SLMs for Coding Agents is positioned in show hn launch, a segment that rewards a clear, focused offering.

Its current recorded stage is Show HN launch.

Website: www.autohand.ai/updates/announcing-fantail-models