Thai LLM Leaderboard v0 — single-protocol scores from arXiv:2504.01789

README.md

@@ -1,10 +1,26 @@
 ---
-title: Thai Llm Leaderboard
-emoji: 🏃
-colorFrom: gray
-colorTo: green
+title: Thai LLM Leaderboard
+emoji: 📊
+colorFrom: blue
+colorTo: indigo
 sdk: static
-pinned: false
+pinned: true
+short_description: Thai LLMs benchmarked under one identical protocol
 ---
 
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+# Thai LLM Leaderboard
+
+Scores for Thai large language models evaluated under **one identical protocol** — same
+prompts, same harness, every model.
+
+- **v0 (current):** all scores from the single SkyThought evaluation run published in the
+  [OpenThaiGPT 1.6 & R1 technical report](https://arxiv.org/abs/2504.01789) (April 2025).
+- **v1 (planned):** fresh re-runs including ThaiLLM consortium models, newer Typhoon
+  releases, Pathumma updates, THaLLE, and closed APIs (GPT, Gemini).
+
+Reproduce or submit a model: [openthaigpt_eval](https://github.com/OpenThaiGPT/openthaigpt_eval)
+· [Discord](https://discord.gg/7KDdKkBGUs)
+
+Maintained by [OpenThai](https://openthai.aieat.or.th) (AIEAT · iApp Technology).
+The best score in each column is computed, not chosen — our models do not win every
+column, and that is the point.