Trim org card to identity-only (no code, no launch details)

README.md

@@ -9,28 +9,6 @@ pinned: false
 
 # Zero One Research
 
-**Indie AI research lab in Bratislava, EU.** We build small, calibrated foundation models for tabular data — the structured rows that don't fit a chatbot.
+Indie AI research lab in Bratislava, EU. We build small, calibrated foundation models for tabular data — the structured rows that don't fit a chatbot.
 
-## PredictLM v1
-
-Two open-weight tabular foundation models, Apache-2.0:
-
-- [`predictlm-mini-13m`](https://huggingface.co/zerooneresearch/predictlm-mini-13m) — 13.5M parameters, 54 MB. Distilled from Base. CPU / edge.
-- [`predictlm-base-26m`](https://huggingface.co/zerooneresearch/predictlm-base-26m) — 26.2M parameters, 105 MB. Highest-accuracy model in the family.
-
-**Headline**: 0.751 classification accuracy / 0.609 regression R² on a locked 25-dataset OpenML benchmark via the published Duo + TTT (test-time training) recipe. Built by one person on a Mac Studio in two weeks. Negative results documented alongside the wins.
-
-```python
-from predictlm import PredictLM
-model = PredictLM.from_pretrained("zerooneresearch/predictlm-mini-13m")
-preds = model.fit(X_train, y_train).predict(X_test)
-```
-
-## Open by default — with one caveat
-
-v1 weights and research are open under Apache-2.0. Future premium models may follow a Mistral-style split — community-tier open, premium hosted-only. We'll be explicit on each release.
-
-## Companion artifacts
-
-- **MCP server** for LLM agents (Claude Desktop, Cursor, Continue): `pip install predictlm-mcp` ([source](https://github.com/matej-01RAI/predictlm-mcp))
-- **Python package**: [`predictlm`](https://pypi.org/project/predictlm/) on PyPI
+Our first release, **PredictLM v1**, is open under Apache-2.0. Future premium models may follow a Mistral-style split — community-tier open, premium hosted-only. We'll be explicit about which tier each release is on.