researchclaw-hybrid-llm-haiku-not-gpt4o-mini

AutoResearchClaw’s internal pipeline uses a hybrid LLM strategy: Claude Opus 4.6 (via ACP) for the 8 critical reasoning stages (hypothesis gen, paper writing, peer review) and Claude Haiku 4.5 for the 15 routine stages. GPT-4o-mini was initially proposed for routine stages but rejected — stay within the Anthropic ecosystem to preserve prompt cache sharing and avoid external API dependencies. AutoResearchClaw natively supports primary + fallback model config.

Related

• autoresearchclaw-hybrid-llm-routing-opus-haiku
• hybrid-llm-model-pattern-opus-haiku-pipeline-stages
• hybrid-opus-haiku-for-researchclaw-stay-anthropic
• researchclaw-hybrid-llm-opus-haiku-stage-split
• haiku-4-5-for-routine-researchclaw-stages-not-gpt4o-mini
• superpowers-plugin-v5-all-14-skills-already-installed