Releases·Model Release
BAAI releases AREX-2, a 27B long-horizon agent model built on Qwen3.8 with test-time self-improvement
The Beijing Academy of Artificial Intelligence (BAAI) released AREX-2, a 27B-parameter agent model on Hugging Face. It improves solutions across multiple test-time rounds via a propose, measure, reflect, revise loop, trained on ML and algorithmic-programming tasks with verifiable feedback. The learned self-improvement behavior transfers to deep research without new search trajectories.
100 engagement1 sourceWed, Sep 30, 2026, 08:21 AM
AI Overview
- BAAI's AREX-2 is a dense Qwen3.8-compatible multimodal agent model with 27B parameters and 262,144-token context.
- AREX-2 iterates via a propose-measure-reflect-revise loop, reading scores, logs, errors, and timings to decide revisions.
- Trained on ML and algorithmic-programming tasks with verifiable feedback plus AREX deep-research data, its self-improvement transfers to deep research without new search trajectories.
- GGUF quantizations are available from mradermacher on Hugging Face.
Entities
Related
Developments·Paper
AREX paper proposes recursively self-improving agents for deep research
A new arXiv paper introduces AREX, a family of recursively self-improving (RSI) deep research agents that exploit the discovery–verification asymmetry in constraint-satisfaction problems. AREX alternates between verifying intermediate results and using the partially verified state to guide further refinement, aiming to reduce the cost of finding answers that satisfy multiple constraints.
A new arXiv paper introduces AREX, a family of recursively self-improving (RSI) deep research agents that exploit the discovery–verification asymmetry in constraint-satisfaction problems. AREX alternates between verifying intermediate results and using the partially verified state to guide further refinement, aiming to reduce the cost of finding answers that satisfy multiple constraints.
0 engagement1 sourcearxivThu, Jul 23, 2026, 04:05 PM
Releases·Model Release
Tencent releases ContextPilot models for proactive agent context management
Tencent released ContextPilot, a proactive context-management framework for long-horizon language-model agents, as open-weight checkpoints in 14B, 8B, and E4B sizes. The 14B model is a Qwen3-14B checkpoint trained to plan, maintain long-term memory, and offload less useful context while agents keep reasoning and using tools. Weights are on Hugging Face with an accompanying paper and code repository, giving practitioners a drop-in approach to context management in agentic systems.
Tencent released ContextPilot, a proactive context-management framework for long-horizon language-model agents, as open-weight checkpoints in 14B, 8B, and E4B sizes. The 14B model is a Qwen3-14B checkpoint trained to plan, maintain long-term memory, and offload less useful context while agents keep reasoning and using tools. Weights are on Hugging Face with an accompanying paper and code repository, giving practitioners a drop-in approach to context management in agentic systems.
6 engagement1 sourceredditMon, Aug 31, 2026, 08:56 AM
Releases·Model Release
XYZ AI Lab releases XYZ-Aquila-mini, an open-weight Deep Search agent
XYZ AI Lab released XYZ-Aquila-mini, an open-weight Deep Search agent post-trained from Qwen3.6-35B-A3B via an AI4AI pipeline. The model is a thinking model designed for bounded-exploration search tasks, with human-defined constraints and AI-driven failure diagnosis.
XYZ AI Lab released XYZ-Aquila-mini, an open-weight Deep Search agent post-trained from Qwen3.6-35B-A3B via an AI4AI pipeline. The model is a thinking model designed for bounded-exploration search tasks, with human-defined constraints and AI-driven failure diagnosis.
10 engagement1 sourceredditMon, Jul 27, 2026, 12:00 PM
Developments·Paper
Beyond Final Scores: A Systematic Evaluation of Agents for Long-Horizon AI Research and Development
A new arXiv paper presents a systematic evaluation of seven frontier models on 36 long-horizon tasks, using rule-based metrics to characterize within-run behavior beyond final scores. The framework tracks Solution Framing, Execution, and Feedback to reveal where progress is gained or lost and whether accumulated experience improves later decisions.
A new arXiv paper presents a systematic evaluation of seven frontier models on 36 long-horizon tasks, using rule-based metrics to characterize within-run behavior beyond final scores. The framework tracks Solution Framing, Execution, and Feedback to reveal where progress is gained or lost and whether accumulated experience improves later decisions.
0 engagement1 sourcearxivThu, Aug 13, 2026, 04:11 PM
Developments·Paper
APEx distills agent procedural experience for adaptive deep research question answering
An arXiv paper introduces APEx, a method for distilling agent procedural experience to improve deep research question answering. It targets the gap where existing approaches either retrieve verbose task-specific traces that burden decision-making or distill procedural skills decoupled from downstream policy adaptation.
An arXiv paper introduces APEx, a method for distilling agent procedural experience to improve deep research question answering. It targets the gap where existing approaches either retrieve verbose task-specific traces that burden decision-making or distill procedural skills decoupled from downstream policy adaptation.
0 engagement1 sourcearxivWed, Sep 2, 2026, 08:00 AM