Files
ai-podcast/backend
luke fcefabdaee Expand style-matched routing to 10 models for maximum caller variety
- Grok 4.1 Fast: high_energy, bragger, comedian, small_town_gossip
- Grok 4 Full: confrontational (needs deep reasoning for arguments)
- Claude Sonnet 4.6: quiet_nervous, emotional (genuine vulnerability)
- Kimi K2: sweet_earnest (warm, creative, different texture than Claude)
- Mistral Large: deadpan, mysterious (dry, precise)
- DeepSeek Chat: angry_venting (raw, unfiltered rage)
- DeepSeek R1 Distill: oversharer, conspiracy (commits fully, no hedging)
- Qwen: storyteller, rambling (loves tangents and detail)
- Gemini 2.5 Pro: know_it_all (pedantic, cites sources)
- Llama 3.3 70B: world_weary, reluctant, first_time (casual, natural)

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-21 02:40:09 -06:00
..