GRM-2.7-Mythos is an advanced closed 27B parameter model specialized in cybersecurity. Trained exclusively on verified security traces, it delivers performance that surpasses expectations for its scale — demonstrating strong capabilities in threat analysis, vulnerability detection, and security reasoning.
Training
| Stage | Details |
|---|---|
| SFT | 2K filtered cs.CR papers + real CVE data |
| RLVR | GitHub repos with past vulnerabilities, paired with a verifier model checking outputs against ground truth |
Contact
This model is closed-weight and not publicly hosted. If you'd like access, please reach out to dedeprogames_official on Discord with a brief explanation of your use case.
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support
Model tree for OrionLLM/GRM-2.7-Mythos
Base model
Qwen/Qwen3.6-27B