FOCUS uses selective branching to train long-horizon agent critics
At Qwen3.5-9B, held-out accuracy rises 7.2 points on BrowseComp-Plus and 4.1 on DeonticBench versus the strongest baseline.
Stories per day · 30 days
1 RSS
Relatedqwen2reinforcement-learning2agent-experience1agentic-rl1benchmarks1claude1context-compaction1credit-assignment1drex1gpt1language-agents1multi-session1nace-ai1rege1search-agents1subagents1test-time-scaling1
11 stories, newest first · one card per story
Stories tagged #browsecomp
Nace AI releases Drex search system for AI agents
The headline cites 92.05% on BrowseComp-Plus using 12.6 search calls; the article body was not available.
ReGE turns recorded agent experience into guidance for a retry
ReGE-Syn raises accuracy 9.9 points for GPT and 8.8 for Claude while cutting guidance-construction tokens versus adapted AggAgent.
Mosaic: Multi-Session Agentic Reinforcement Learning
Mosaic improves multi-session agentic RL by crediting session contributions instead of broadcasting trajectory advantages.
SearchJev: A Fast and Calibrated System-1 Model for Search Agents
SearchJev makes calibrated search decisions 5.2× faster than Qwen3.5 and lifts BrowseComp-Plus accuracy from 45% to 54%.
Context Language Models
Context Language Models edit their own context file, beating external managers on agent benchmarks while cutting FLOPs.
Adaptive Consistency Graph for Long-Horizon Agents
Adaptive Consistency Graph raises GPT-5.6-luna long-horizon success from 44.5% to 50.2%, led by BrowseComp-Plus.
IterSynth: Rethinking Deep Search Agents via Role-Decoupled Iterative Synthesis
IterSynth decouples planning and synthesis for deep-search agents, and an 8B model scores 50.7.
Grow the Harness, Not the Context: From Strategy-Free Scaffolds to Reusable Specialist Agents
Growing Harness converts recurring agent control into reusable code, cutting LLM calls by 76 to 92 percent.
Iris-mini and Iris-pro are the strongest open-weight search agents in their class
Chinese lab AllSpark releases Iris-mini (35B) and Iris-pro (397B) open-weight search agents claiming best-in-class results on BrowseComp and other research benchmarks.
3
Can LLMs Engineer Their Own Agent Harness? ByteDance Seed’s HarnessDev Says Only 34 of 64 Changes Generalize
ByteDance Seed's HarnessDev benchmark finds LLM-built agent harnesses trail human engineering on code and search, with only 34 of 64 revisions generalizing.
9