AI Paper Atlas

Research map / Large language models / Agents, code and math

Scaling Agents via Continual Pre-training

Liangcai Su, Zhen Zhang, Guangyu Li, Zhuo Chen, Chenxi Wang, Maojia Song, Xinyu Wang, Kuan Li, Jialong Wu, Xuanzhong Chen, Zile Qiao, Zhongwang Zhang and 14 more

ICLR 2026 · Large language models › Agents, code and math · approach: Long-horizon LLM agents · 39 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add agentic continual pre-trainingQwen3-30B-A3B-Base · BrowseComp-en (SFT-A)Pass@1: 26.9 -> 31.4Improved
add agentic continual pre-trainingQwen3-30B-A3B-Base · BrowseComp-en (SFT-C)Pass@1: 24.5 -> 38.8Improved
add agentic continual pre-trainingQwen3-30B-A3B-Base · HLE (SFT-C)Pass@1: 27.9 -> 28.9Improved
add agentic continual pre-trainingQwen3-30B-A3B-Base · GAIA (SFT-B)Pass@1: 71.8 -> 72.8Improved
add second long-context stage to single-stage CPT (ablation)AgentFounder · BrowseComp-enPass@1: 31.4 -> 35.5Improved
add second long-context stage to single-stage CPT (ablation)AgentFounder · BrowseComp-zhPass@3: 50.5 -> 58.5Improved
add second long-context stage to single-stage CPT (ablation)AgentFounder · GAIAPass@3: 81.6 -> 82.5Improved
add FAS data for CPT (ablation)Qwen3-30B-A3B-Base · BrowseComp-zhPass@1: 29.8 -> 37.0Improved
add HAS data to FAS data (ablation)AgentFounder CPT · BrowseComp-zhPass@1: 37.0 -> 40.1Improved
add HAS data to FAS data (ablation)AgentFounder CPT · GAIAPass@1: 72.8 -> 69.9Hurt
add HAS data to FAS data (ablation)AgentFounder CPT · BrowseComp-enPass@1: 31.4 -> 31.4No clear effect
scale model size from 1B to 4B to 30B (ablation)AgentFounder · deep research benchmarks (average)accuracy: 20.4% -> 32.7% -> 48.9%Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy