hybrid-llm-config-opus-critical-stages-haiku-routine

AutoResearchClaw supports primary + fallback model configuration natively. For Oracle integration: use Opus 4.6 for the 8 critical reasoning stages (stages 2, 7, 8, 9, 14, 15, 17, 18 — hypothesis generation, literature synthesis, experiment design, paper writing, peer review simulation) and Haiku 4.5 for the 15 routine stages (topic init, resource planning, export, status checks). Stay within the Anthropic ecosystem rather than mixing in GPT-4o-mini — avoids credential sprawl and keeps quality consistent. AutoResearchClaw calls models via ACP delegating through Claude Code.

Related

• researchclaw-hybrid-llm-haiku45-for-routine-stages
• hybrid-llm-opus-critical-haiku-routine-anthropic-only
• hybrid-llm-routing-opus-critical-haiku-routine-for-researchc
• hybrid-llm-model-pattern-opus-haiku-pipeline-stages
• autoresearchclaw-hybrid-llm-routing-opus-haiku
• hybrid-llm-routing-opus-critical-haiku-routine-in-pipeline
• oracle-researchclaw-hybrid-llm-opus-haiku-split
• hybrid-llm-strategy-for-long-research-pipelines
• hybrid-opus-haiku-llm-config-for-autoresearchclaw-stages