hybrid-llm-routing-opus-critical-haiku-routine-in-pipeline

For multi-stage research pipelines, use Opus 4.6 only for the 8 critical reasoning stages (hypothesis generation, paper writing, peer review simulation) and Haiku 4.5 for the 15 routine stages (topic init, resource planning, export, formatting). AutoResearchClaw natively supports primary + fallback model configuration, making this trivially configurable. Stay within the Anthropic ecosystem rather than routing routine stages to GPT-4o-mini to avoid cross-provider complexity.