Add card (28 Oct 2026)

README.md

@@ -0,0 +1,60 @@
+---
+pipeline_tag: text-generation
+license: cc-by-4.0
+tags:
+  - owl-alpha
+  - longcat-2.0
+  - stealth-model
+  - stealthmark
+  - coding
+  - tool-use
+  - long-context
+thumbnail: https://stealthmodels.com/owl-alpha/assets/owl-head-5.webp
+new_version: meituan-longcat/LongCat-2.0
+---
+
+<table>
+  <tr>
+    <td width="65%" valign="middle">
+      <h1>Owl Alpha</h1>
+      <p><em>You have my attention.</em></p>
+      <p><strong>Revealed as LongCat-2.0 by Meituan.</strong></p>
+      <p>Owl Alpha pairs native tool use with a one-million-token context window, and Meituan has since released the weights under the MIT license.</p>
+      <p><a href="https://stealthmodels.com/owl-alpha/"><strong>Meet Owl Alpha ↗</strong></a></p>
+    </td>
+    <td width="35%" align="center" valign="middle">
+      <a href="https://stealthmodels.com/owl-alpha/">
+        <img src="https://stealthmodels.com/owl-alpha/assets/owl-head-5.webp" width="260" alt="A cloaked owl raises her head.">
+      </a>
+    </td>
+  </tr>
+</table>
+
+> Independent benchmark card by [Stealth Models](https://stealthmodels.com/): our test results and findings for this model.
+
+## Facts
+| First seen | Revealed | Developer | Context | Input | Tool calls | Speed in our tests | Access |
+|---|---|---|---|---|---|---|---|
+| 28 Apr 2026 | 29 Jun 2026 | Meituan | 1,048,576 tokens | Text | Yes, native | about 86 tok/s | Stealth access ended; available as LongCat-2.0 |
+
+## StealthMark results
+| Benchmark score | General Intelligence | Visual Intelligence |
+|---|---|---|
+| 25.5 | 17.2 | 34.4 |
+
+Its best placing is Pelican, subject construction and spatial relationships: 79.0, third. Among the tasks that count toward the benchmark score, its highest result is [Selfie](https://stealthmodels.com/benchmarks/models/owl-alpha/selfie), visual construction and creative interpretation, at 61.6 out of 100.
+StealthMark 1.8.8, results read 28 October 2026. The tested version is the full June release, listed as LongCat-2.0. [Full results](https://stealthmodels.com/benchmarks/models/owl-alpha) · [On the map](https://stealthmodels.com/benchmarks/map)
+
+<img src="https://stealthmodels.com/benchmarks/media/ad3b9ff625b0748a.png" width="360" alt="LongCat-2.0's Selfie result, its highest counted benchmark task.">
+
+## How it was revealed
+- **29 Jun 2026:** Meituan identified LongCat-2.0 as the model behind Owl Alpha, in an [announcement](https://www.linkedin.com/posts/meituan_meituan-longcat-llm-activity-7477722701768499200-4qrk). The full LongCat-2.0 launched on 30 Jun, and the weights followed under the MIT license on 12 Jul ([Meituan's notes](https://tech.meituan.com/2026/07/12/LongCat-2.0-Open-source.html)).
+- **30 Apr 2026:** Community guesses named GLM 5.2 and LongCat-2.0-Preview. Meituan's own description of the Preview listed tool calling, long context and coding, which made it a plausible candidate.
+- **28 Apr 2026:** Owl Alpha first appeared with free access, a 1,048,576-token context and no developer's name.
+
+## Run it
+Official weights [meituan-longcat/LongCat-2.0](https://huggingface.co/meituan-longcat/LongCat-2.0) · license: MIT, [read it](https://huggingface.co/meituan-longcat/LongCat-2.0/blob/main/LICENSE) · Local: about 48B active parameters per token, but the full model needs rack-scale hardware, so it is not a local model.
+
+---
+[Model guide](https://stealthmodels.com/owl-alpha/) · [StealthMark](https://stealthmodels.com/stealthmark) · [Play the benchmarks](https://stealthmodels.com/play) · [Follow for new reveals](https://huggingface.co/stealth-model)
+This card and its images: CC BY 4.0, credit stealthmodels.com. The model's weights have their own license. Updated 28 October 2026.