Hyperflow H3 hits #5 on I2V and #3 on video editing at under half Seedance 2.5's price
ArtificialAnlys · x · 2026-10-31
Artificial Analysis published a full evaluation of Hyperflow H3, Video Rebirth's 8-step open-weights LoRA built on MiniMax H3, now supporting image-to-video and video editing.
- Ranks #5 on AA-Video-I2V v2.1 and #3 on AA-Video-Editing v1.1, beating its base MiniMax H3 (768p) on editing
- Priced at $9.00/min for I2V and $18.00/min for editing (1088p) — under half of Dreamina Seedance 2.5 ($34.12 and $40.82/min)
- Strongest in Live-Action Film (#2, only Seedance 2.5 higher) and Multi-Scene & Narrative; biggest gains over the base model in spatio-temporal consistency and audio sync; weakest in human anatomy and lip sync
- In editing, best at visual effects (only Wan 3.0 higher) and sound/speech editing; trails its base on physics & world simulation
More from Models
- SenseNova-U1.5-8B hits 125 tok/s on M5 Ultra in llama.cpp and spots a $1 receipt error — Binary_orchid · 2026-10-31
- Pundit: Anthropic killing 5-hour limit plus Fable launch makes it unstoppable — iruletheworldmo · 2026-10-31
- ChatGPT Daybreak Blue sessions may train OpenAI models even with training opt-out off — littmath · 2026-10-31
- Claude refuses to draw its own logo over copyright concerns — 141_1337 · 2026-10-31
- AI cracks a 27-year-old pillar of computer science, OpenAI and Anthropic claim — Fcking_Chuck · 2026-10-31
- Perplexity early evals: Gemini 4 Carbon tops WANDR benchmark for long-form research — AravSrinivas · 2026-10-31