Add model card

README.md

@@ -0,0 +1,116 @@
+---
+language:
+  - en
+license: apache-2.0
+library_name: openai-compatible
+pipeline_tag: text-generation
+tags:
+  - stealth-skeleton
+  - mathematical-reasoning
+  - conformal-prediction
+  - optimal-transport
+  - tool-calling
+  - code-generation
+  - livebench
+inference: false
+---
+
+# stealth-skeleton/skeleton-alpha
+
+**Stealth-Skeleton Research (`stealth-skeleton`)**
+
+> **API-only.** This repository is the official Model Card & Documentation Hub for the `stealth-skeleton/skeleton-alpha` endpoint on OpenRouter. It does **not** contain downloadable model weights. The Apache-2.0 license applies to the documentation and configuration files in this repository.
+
+**skeleton-alpha** is a free crowd-RLHF preview: an execution-verified coding, math and tool-calling model from the `stealth-skeleton` family, which targets sovereign mathematical, spectral-operator, conformal-prediction and execution-verified coding/reasoning workloads.
+
+| Property | Value |
+|---|---|
+| Context window | 128K tokens |
+| Max output | 16K tokens |
+| Pricing | Free (community research preview) |
+| Tool calling | Auto-Exacto grammar healing (>95%, self-reported) |
+| Interface | OpenAI-compatible Chat Completions |
+
+## Architecture & Capabilities
+
+- **Execution-verified mathematical operator synthesis**: generated operators are checked by running them.
+- **Sinkhorn-Knopp optimal transport**: L1 marginal residual 1.39e-16 at 40 iterations, epsilon = 0.15.
+- **Frisch-Waugh-Lovell (FWL) orthogonal projection**: residual < 1e-15.
+- **Romano Conformalized Quantile Regression (CQR)**: 90.35% empirical out-of-sample coverage, 0.2337 Normal CRPS.
+- **36th-order Hou-Li spectral filtering**.
+- **Auto-Exacto tool-calling grammar healing**: malformed tool-call JSON is repaired before it is returned.
+
+All figures above are self-reported from internal tests.
+
+## Benchmarks: Internal 3-Pass Evaluation Suite (Self-Reported)
+
+These results come from our internal 3-pass evaluation suite. They have not been independently verified and may differ from official leaderboard runs.
+
+| Suite | Score |
+|---|---|
+| MATH-500 | 92.00% |
+| GSM8K | 96.00% |
+| HumanEval | 92.00% |
+| GPQA-Diamond | 73.33% |
+| MMLU-Pro | 86.00% |
+| Arena-Hard | 85.00% |
+| ForecastBench (Proper Brier Skill) | 78.52% (0.0969 Brier) |
+| **Composite (mean of 7)** | **86.12 / 100** |
+
+## Usage
+
+Model ID: `stealth-skeleton/skeleton-alpha` via OpenRouter, once provider routing is active for your account.
+
+### Python (OpenAI SDK)
+
+```python
+from openai import OpenAI
+
+client = OpenAI(
+    base_url="https://openrouter.ai/api/v1",
+    api_key="<YOUR_OPENROUTER_API_KEY>",
+)
+
+resp = client.chat.completions.create(
+    model="stealth-skeleton/skeleton-alpha",
+    messages=[{"role": "user", "content": "Prove that the sum of two even integers is even."}],
+)
+print(resp.choices[0].message.content)
+```
+
+### curl
+
+```bash
+curl https://openrouter.ai/api/v1/chat/completions \
+  -H "Authorization: Bearer $OPENROUTER_API_KEY" \
+  -H "Content-Type: application/json" \
+  -d '{
+    "model": "stealth-skeleton/skeleton-alpha",
+    "messages": [{"role": "user", "content": "Write a Python function for the Sinkhorn-Knopp algorithm."}]
+  }'
+```
+
+### sst/opencode
+
+```json
+{
+  "$schema": "https://opencode.ai/config.json",
+  "provider": {
+    "openrouter": {
+      "options": { "apiKey": "<YOUR_OPENROUTER_API_KEY>" },
+      "models": { "stealth-skeleton/skeleton-alpha": {} }
+    }
+  },
+  "model": "openrouter/stealth-skeleton/skeleton-alpha"
+}
+```
+
+## Data & Privacy
+
+`skeleton-alpha` is a **free community research preview**. Anonymized mathematical and tool-calling traces may be evaluated by our automated execution verifier for DPO/GRPO alignment. Do not submit confidential, personal or proprietary data to this endpoint. For zero-retention use, see `stealth-skeleton/skeleton-v1`.
+
+## Limitations
+
+- Preview model: behavior, limits and availability may change without notice.
+- Self-reported benchmarks only; validate on your own workloads.
+- Not intended for safety-critical, medical, legal or financial decisions.