LTXV 13B Distilled – Generate 5s Videos in Under 10s
Hey HN, after our 13B release we've been working on a faster version of our open-source video model and we're excited to share it. We started with a 13B base model that already had competitive generation speeds (e.g. 55s for a 5s video on an H100 — faster than any other model out there). But we wanted to push it further to allow everyone to quickly iterate over their video generations. So we built a distilled version focused on speed without sacrificing temporal or spatial coherence. With the Distilled model, you can now generate 5-second 720p videos in about 9.5 seconds on an H100, and…
In plain words
LTXV 13B Distilled is an open-source video generation model designed to create 5-second 720p videos quickly. It generates videos in approximately 9.5 seconds on an H100 GPU and around 1.5 minutes on consumer GPUs like the RTX 5090, prioritizing speed while maintaining temporal and spatial coherence. The distilled model is interoperable with the base 13B model, allowing developers to mix and match them in rendering pipelines for flexible workflows. It's intended for developers and creators who need fast video iteration and experimentation.
written from the facts on this page · September 2026
From the sources
In the maker’s words, at launch
Hey HN, after our 13B release we've been working on a faster version of our open-source video model and we're excited to share it. We started with a 13B base model that already had competitive generation speeds (e.g. 55s for a 5s video on an H100 — faster than any other model out there). But we wanted to push it further to allow everyone to quickly iterate over their video generations. So we built a distilled version focused on speed without sacrificing temporal or spatial coherence. With the Distilled model, you can now generate 5-second 720p videos in about 9.5 seconds on an H100, and around 1.5 minutes on a consumer GPU like an RTX 5090. Because the base and distilled models are interoperable, you can mix and match them in a single rendering pipeline. That gives you three modes: - Distilled Pipeline: Fastest. Uses just 4–8 steps end to end. Ideal for rapid iteration and experimentation (~9.5 seconds on H100, ~1.5 min on RTX 5090). - Mixed Pipeline: Starts with the base model to capture accurate motion, physics, and detail, then switches to distilled at higher resolutions for faster rendering. A good middle ground. (~20s on H100, ~2.5 min on RTX 5090) - Base Pipeline: Full-fidelity generation from start to finish. Best quality, best for final renders. (~43s on H100) All pipelines are compatible with our multiscale rendering system, which allows you to first render a lower resolution video and upscale it as needed. The project is open source and available on GitHub (https://github.com/Lightricks/LTX-Video/) and Hugging Face (https://huggingface.co/Lightricks/LTX-Video-0.9.7-distilled). You can also enjoy our Comfy integration here (https://github.com/Lightricks/ComfyUI-LTXVideo) and our open source LTXV trainer (https://github.com/Lightricks/LTX-Video-Trainer). Would love to hear your thoughts — especially around performance, integration into your own tools or workflows, or any creative uses you're exploring. We're actively working on tuning and adding new configurations, and early feedback is super helpful.
Does the same job
all alternatives →
- TTText-to-video model from scratch (2 brothers, 2 years, 2B params)Jan 2026 · huggingface.co · ▲158
Writeup (includes good/bad sample generations): https://www.linum.ai/field-notes/launch-linum-v2 We're Sahil and Manu, two brothers who spent the last 2 years training text-to-video models from scratch. Today we're releasing them under Apache 2.0. These are 2B param models capable of generating 2-5 seconds of footage at either 360p or 720p. In terms of model size, the closest comparison is Alibaba's Wan 2.1 1.3B. From our testing, we get significantly better motion capture and aesthetics. We're not claiming to have reached the frontier. For us, this is a stepping…
- DDDistilling DeepSeek into GPT-OSS doesn't transfer censorship. Try itJul 2026 · ctgt.ai · ▲170
We recently used DeepSeek V4 Flash as a teacher for finance tasks with GPT-OSS-120B. Distillation works well on this problem. At a constrained 8k token budget, our self-distilled 120B scores 83.61% on FinanceReasoning, above Kimi K3 (81.93%) and Inkling (65.13%). We released the 20B open weights. With V4 as the teacher though, we realized it would be timely to measure if the censorship characteristic of it transferred to the distilled version of the base model. tl;dr it didn't, the teacher answered politically sensitive questions 7 SDs differently than expected, but the distilled model's…
More dev tools this month
the category →
The first open-source price index for GPU compute
Dev tools · 10d ago · getcomputable.com
OpenTrailPaper is open-source bike computer firmware for the LilyGO T5S3 4.7" E-Paper PRO. It supports offline maps, GPX routes, FIT recording and Bluetooth sensors.
Dev tools · 2d ago · opentrailpaper.com
Open-source GTM skills for technical founders
Dev tools · 29d ago · gtmcofounder.com
Launched alongside, May 2025
the whole month →
- C9
Life & fun · 2025 · felixrieseberg.github.io
