H3 NSFW Animus Uncensored run locally. - 131k context. - Refusal rate: 8/100. - KL divergence: 0.0200 That KL is extremely low. Most abliterations trade writing quality for compliance. This one barely moved the probability mass. Not for all audiences. Treat it like an uncensored creative model, not a general assistant. -

H3 🐰💋👠 N$FW LoRA -PinkFluffyBunny lora that truly brings the style and pose -Maximum results achieved at 0.5 str on pruned int8 model. -Use the h3_fl2va_pruned_int8_convrot.safetensors model. Alpha quality so temper expectations 👇😉

@h3x_tech 🤝 @SailGP We’re proud to power the state-of-the-art jet propulsion system that will help SailGP's fleet of identical F50 catamarans foil in light wind conditions - currently under development for seasons ahead. Developed by HamiltonJet, our electric motor drives the Pod-Jet system providing greater certainty of world-class racing for both athletes and fans. Check out SailGP's latest video for more details.

MiniMax H3 Highspeed and H3 Lite from @MiniMax_AI are now available on Merge Gateway. H3 Lite costs half what H3 Max does at 768p ($0.04/sec) and $0.025/sec at 480p, with 5-second clips back in 8 to 9 seconds. H3 Highspeed is 25% under H3 Max at $0.06/sec. Check it out:

MiniMax H3 (Hailuo 3) video on 2× DGX Spark 🎬 • 5 s 1344×768 clip with audio: 643 s (one Spark: 1399 s) • 10 s: 1690 s, 15 s: 3080 s (one Spark: out of memory) • 20 s: 6740 s, beyond the model's 15 s spec and still coherent to the last frame (clip below, one continuous generation, seed 1101, 20 steps) • Text-encoder TP2 + Ulysses SP2 over one ConnectX-7 cable (RoCE, 19.5 GB/s) • Online FP8 DiT, cuDNN attention, 24 fps, native audio • Frames leave the GPU as uint8, VAE tiles decoded on both GPUs: same bytes, 15 s fits with 25 GB to spare on each host • H3-Base open weights (768p), vLLM-Omni, no Ray and no changes to the vLLM-Omni tree vllm serve cannot span two hosts, so the recipe adds a cross-host executor: rank 1 runs on the second Spark and is driven over the link through vLLM's own message queues, behind the unchanged single-node API. Quality checks, memory traces and the open levers are in the repo. Full recipe, measurements and credits: