AGENTS.md@docs/models git:20261004.f5b8c87A

Immutable. This exact content never changes and is served at /api/v1/blob/0202051e1513968b.

# DOX — docs/models

AI-model catalog. Generated by `packages/video-transcription/.pi/skills/youtube-srt/scripts/build-catalog.py` from AI Search YouTube transcripts (57 videos, 2026-04-05 → 2026-10-04). Generated files: do NOT hand-edit; regen cmd in `README.md`. Facts = presenter claims, unverified. One row per file, path-alphabetical.

| File | Purpose |
|------|---------|
| `3d.md` | Generated. 3D generation + reconstruction models (45). |
| `README.md` | Generated. Catalog index: by-function table, by-hardware link, data schema, regen cmd. |
| `agent.md` | Generated. Agents + AI tools (52). |
| `asr-speech.md` | Generated. Speech recognition (10). |
| `avatar-lipsync.md` | Generated. Avatars + lip-sync (9). |
| `coding.md` | Generated. Coding models & agents (16). |
| `embedding-other.md` | Generated. Embeddings + other (17). |
| `image-edit.md` | Generated. Image editing (20). |
| `image-gen.md` | Generated. Image generation (50). |
| `llm.md` | Generated. LLMs — general chat / reasoning (94). |
| `local-hardware.md` | Generated. Local-runnable models by VRAM tier T0→T5 + unspecified; `est.` = tier from checkpoint size. |
| `models.jsonl` | Generated data source. One merged model per line; fields incl. `functions[]`, `open_weights`, `vram_tier`, `strengths`, `benchmarks`, `mentions[]`. |
| `multimodal-omni.md` | Generated. Multimodal / omni models (30). |
| `music.md` | Generated. Music generation (17). |
| `picks.md` | Hand-curated. Best-for-task + best-per-hardware shortlist. Never written by `build-catalog.py`; refresh by hand after each regen. |
| `realtime-voice.md` | Generated. Realtime voice assistants (11). |
| `robotics.md` | Generated. Robotics (70). |
| `science-medical.md` | Generated. Science & medical (21). |
| `tts-voice.md` | Generated. Text-to-speech & voice cloning (17). |
| `video-edit.md` | Generated. Video editing (21). |
| `video-gen.md` | Generated. Video generation (48). |
| `world-model.md` | Generated. World models & interactive worlds (41). |