AI
Marathoner: Ultra-Long-Horizon Autonomous Intelligence
Publication: Hugging Face Daily Papers
Author: Not available in collected metadata
Source published: 27/09/2026, 20:00:00 UTC
Briefing created: 30/09/2026, 04:45:35 UTC
Based on a feed excerpt; full article content has not been acquired. Source evidence retrieved 30/09/2026, 03:18:25 UTC.
Briefings
Marathoner is an autonomous agentic model developed in 2026 that enables ultra-long-horizon execution by synthesizing challenging tasks from GitHub PRs and using Multi-Task Chaining. The model employs rejection sampling finetuning with a strong teacher model and a novel Later Stage Bonus Reward to acquire genuine capabilities. Evaluated on five benchmarks, Marathoner consistently outperformed base models and proprietary strong models, demonstrating the ability to work for over 10 hours and execute over 1000 tool calls on difficult tasks.
Key points
- Marathoner is an autonomous agentic model introduced in 2026 that uses a comprehensive post-training pipeline to instill ultra-long-horizon execution capabilities.
- The model synthesizes challenging task-level data by leveraging major release PRs containing 1000+ lines of new code from diverse GitHub repositories.
Context and limitations
Evaluation was conducted on five benchmarks containing ultra-long-horizon tasks, and results are subject to the snapshot limitations of the research period.
Generated by Bittrees-hosted qwen35-2b sha256:aaf42c8b7c3c on 30/09/2026, 04:45:35 UTC. Based on publisher-provided feed evidence; may not cover the full original article. Check the source for details.
View archived briefing on IPFS
bafybeian4zxxeaad6du42lxtnn7laij52pn5ytfd42vvrx6zvzazygyqie