autoresearchclaw-hybrid-llm-routing-opus-haiku

AutoResearchClaw’s 23 pipeline stages use a hybrid LLM strategy: Claude Opus 4.6 for the 8 critical reasoning stages (hypothesis generation, paper writing, peer review simulation) and Claude Haiku 4.5 for the 15 routine stages (topic init, resource planning, export). Haiku 4.5 was chosen over GPT-4o-mini to stay within the Anthropic ecosystem and avoid cross-vendor API cost/latency. AutoResearchClaw natively supports primary + fallback model config.