Delaying early consensus: ArcticSwarm hits 82.6% on BrowseComp-Plus with gated agent isolation
rohanpaul_ai · x · 2026-09-08
The arXiv paper ArcticSwarm tackles a key failure mode in long-horizon multi-agent research: when agents share partial findings too early, one plausible answer attracts all the search effort before alternatives are tested.
Key design:
- Separating evidence gathering from integration: subagents publish to a shared bulletin board instead of syncing in real time
- Gated isolation: selected search tasks keep their own prior, blocking premature consensus
- Structured review at three commitment boundaries: only confident candidates propagate
Results: 82.6% on the full BrowseComp-Plus with open-weight Qwen 3.5-27B (78.8% without gated isolation, 74.5% without structured review, vs. MiroFlow baseline at 70.6%). On live-web BrowseComp with GPT-5 it reaches 73.6%, well above the reported provider system (54.9%) and MiroFlow (63.4%).
Related event: Delayed Consensus Boosts Multi-Agent Research Performance(2 posts)→
More from coding & agent
- Resin lets an agent build standalone LLM inference engines; vLLM and SGLang optimize DeepSeek kernels differently — AccBalanced · 2026-10-27
- How I let my cloud Grok Bot code on my Mac mini via Tailscale, no public ports — talkaboutdesign · 2026-10-27
- Context layout is a cost decision: one memory edit can bust your entire agent prompt cache — dscape · 2026-10-27
- Alchemy adds effect/Workflow support across Durable Objects, Workflows and Lambda Durable Functions — samgoodwin89 · 2026-10-27
- Alchemy's Lambda Web Functions run plain Node web servers on Lambda, no OpenNext needed — samgoodwin89 · 2026-10-27
- Alchemy makes all ~2,850 AWS bindings work from any cloud, not just S3/Lambda/DynamoDB — samgoodwin89 · 2026-10-27