GLM-5.2 is live. #1 throughput on OpenRouter, pay-per-token on FriendliAI. Try it today ➜
beyoru
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Model Details
Model Provider
beyoru
Model Tree
Input Modalities
Output Modalities
Supported Functionality
| Model | Size | Partial credit | Strict pass |
|---|---|---|---|
| Ornith-1.0-9B | 9B | 22% | 3% |
| seul-preview | 9B | 28% | 7% |
| Qwen3.6-27B | 27B | 34% | 10% |
A custom benchmark also improves performance across all benchmark categories. This serves as a proof of concept (PoC) that reinforcement learning with an environment not only preserves the capabilities of the base model but also improves its overall accuracy.
RLVR training raises partial credit +6 points and more than doubles the strict pass rate over the base, closing a large fraction of the gap to a 3× larger model on the same tasks.
If you use seul in your research or projects, please cite:
bibtex
@misc{seul2026,title = {seul-preview},author = {beyoru},year = {2026},publisher = {Hugging Face},howpublished = {\url{https://huggingface.co/beyoru/seul-preview}}}