---
title: README
sdk: static
pinned: false
---

CloudSurf Software builds **[Surfspace](https://surf.space)** — the AI workspace that turns chats into apps from any device — and trains **CloudSurf** models: small, US-trained, open-weight models for function calling and agentic tool use, released under Apache-2.0.

## Models

- **[CloudSurf-4B-FC](https://huggingface.co/cloudsurf-software/CloudSurf-4B-FC)** — Gemma-4 E4B (4B active / ~8.0B total params) fine-tuned for function calling. BFCL V4 FULL, official harness, 3-seed mean: **55.73** vs 34.81 for the stock base — above the published small-model class bar (51.40). Self-run numbers with full settings disclosure; leaderboard submission open at [gorilla PR #1357](https://github.com/ShishirPatil/gorilla/pull/1357).

## Evidence, not just claims

Every score we publish ships with the raw artifacts to check it:

- [CloudSurf-4B-FC-bfcl-results](https://huggingface.co/datasets/cloudsurf-software/CloudSurf-4B-FC-bfcl-results) — raw BFCL V4 result files for our runs *and* the stock baselines, with per-category scores.
- [tau2-trajectories-cloudsurf-4b-fc](https://huggingface.co/datasets/cloudsurf-software/tau2-trajectories-cloudsurf-4b-fc) — full τ²-bench evaluation trajectories backing our leaderboard submission, open at [tau2-bench PR #480](https://github.com/sierra-research/tau2-bench/pull/480).

Model cards disclose both comparison frames, measured eval-noise bands, and known limitations.

## Links

- Product: [surf.space](https://surf.space)
- Company: [cloudsurf.com](https://cloudsurf.com)