Tavus's Griffin passes the video Turing test: 48% thought it was human in live calls
thetripathi58 · x · 2026-10-08
Tavus launched Griffin, which it calls the first model to pass the video Turing test: 48% of people in live one-minute video calls believed it was a real human, versus a <3% pass rate for previous systems. Griffin listens while you talk and picks up on expressions and pauses, and ranks #1 on NVIDIA's full-duplex AI video benchmark, 37% ahead of the runner-up. Tavus brands it the first Human Interaction Model (HIM).
More from Multimodal
- 3090+3060 local video gen: which model — Wan 2.2, MiniMax H3 or LTX-2? — itspacedev · 2026-10-09
- ComfyUI Face ID workflow troubleshooting: IPAdapter errors one after another — Lanky-Golf-1118 · 2026-10-09
- Mochi: open-source card-based Danbooru tag builder with same-seed A/B previews — bsn28283 · 2026-10-09
- Codex+Tripo toolkit fixes AI 3D models: agent-assisted refinement loop — tinyfool · 2026-10-09
- Trying LTX 2.5 SDR-to-HDR conversion — the results surprised me — ShroakzGaming · 2026-10-09
- Testing LTX 2.5 SDR-to-HDR conversion: results better than expected — ShroakzGaming · 2026-10-09