dfs-large1: First SOTA Cybersecurity Model Built on GLM-5.2 with RL
andreamichi · x · 2026-07-31
depthfirst announced dfs-large1, a new cybersecurity model achieving best-in-class performance in vulnerability detection.
- Architecture & Training: Built on GLM-5.2 and post-trained with reinforcement learning (RL) inside depthfirst's security infrastructure. Large-scale RL compute was powered by Fireworks.
- Evaluation: Achieves state-of-the-art results on depthfirst-bench, a benchmark for long-horizon vulnerability discovery in complex repositories. The team notes training improvements have not plateaued.
- Availability: Currently available in preview within the depthfirst platform.
Related event: depthfirst Releases GLM-5.2-Based Cybersecurity Model dfs-large1(2 posts)→
More from Models
- Tinky Machines Releases Inkling-Small: 276B Parameters with Open Weights — simonguozirui · 2026-07-31
- Using "---" Confuses LLMs and Breaks Role Context — davidad · 2026-07-31
- GPT-5.6 Luna is Cheaper, Smarter, and Faster than Gemini 3.6 Flash — Angaisb_ · 2026-07-31
- LLM Inference Costs Plunge: Token Prices Drop to 1/13th in Four Months — charliermarsh · 2026-07-31
- vLLM Releases Inkling-Small Deployment Guide: Runs on Minimum 180GB VRAM — vllm_project · 2026-07-31
- OpenAI Responds to Benchmark Concerns: GDPval Nearing Saturation — emollick · 2026-07-31