DeepSeek's open-source coding agent DSH statistically ties for first on HF Harness Leaderboard
TheZachMueller · x · 2026-11-11
DSH (DeepSeek Harness), DeepSeek's open-source coding agent, has been added to the Hugging Face Harness Leaderboard, which compares ten coding agents on the same open models.
- Statistically tied for first in all five evaluations: DeepSeek-V4.1-Flash, GLM-5.3-Flash (low/high effort) and Flint
- Shines on DeepSWE (code changes in open-source repos): 2nd of 10 on average, 0.5 points behind NousResearch's Hermes Agent
- Ranks 6th on Terminal-Bench command-line tasks. One habit explains much of the gap: DSH shuts down every background program it started when a task ends. Three Terminal-Bench tasks ask the agent to start a server (e.g. an inference API) that the grader tests afterward — with DSH, the server is already gone.
More from coding & agent
- New Continual-ARC Benchmark Finds Leading AI Agents Fail to Learn From Experience — alifcoder · 2026-11-12
- Azure Day at KubeCon: AKS Runs 22,655 Agents on 65 Nodes in Live Demo — davemccollough · 2026-11-12
- OpenShell tip: expose sandbox web apps to your browser with one command — gethackteam · 2026-11-12
- New Linux-based desktop OS bakes models and agents directly into the operating system — NerdyRodent · 2026-11-12
- Same prompt, three tools: Claude vs ChatGPT vs Figma Make UI face-off — Tegadesigns · 2026-11-12
- DeepLearning.AI launches free short course on tabular foundation models for agents — DeepLearningAI · 2026-11-12