How do you use subagents & multiple agent with local models, and how many?
retained text
Running qwen3.8 27b nvfp4 on vllm at max context only gives around 8 agents with 32k context each. That doesnt seem like much; what use cases do people use multi-agent frameworks and find it helpful for?
document fetched 2026-09-26 00:41 · canonical https://reddit.com/r/LocalLLaMA/comments/1wpxh2g/how_do_you_use_subagents_multiple_agent_with