LTX 2.3
VideoPopularLightricks · 19B · Dense
Open 4K video with native stereo audio — text, image and video-to-video
0
0
2026-03
4K tokens
Use cases
Your hardware
16 GB Reference Desktop
Quantization options
What is quantization?| Quant | Bits | VRAM | Disk | Quality | Status |
|---|---|---|---|---|---|
| Q2_K | 2 | 6.6 GB | 5.8 GB | low | Runs well · ~31 t/s |
| Q3_K_M | 3 | 9 GB | 8.1 GB | moderate | Decent · ~22 t/s |
| Q4_K_M | 4 | 10.2 GB | 9.3 GB | good | Tight fit · ~20 t/s |
| Q5_K_M | 5 | 12.7 GB | 11.6 GB | good | Tight fit · ~16 t/s |
| Q6_K | 6 | 15.1 GB | 13.9 GB | excellent | Barely runs · ~13 t/s |
| Q8_0 | 8 | 20 GB | 18.6 GB | excellent | Barely runs · ~4 t/s |
| F16 | 16 | 39.4 GB | 37.2 GB | lossless | Too heavy · ~1 t/s |
About this model
LTX 2.3 (ltx 2.3) is a 19B video generation architecture released by Lightricks. Unlike autoregressive text LLMs whose KV cache grows with chat turns, local image and video diffusion pipelines allocate memory simultaneously for the transformer backbone, companion text encoders (such as T5/CLIP), and high-resolution VAE latent decoding buffers.
Baseline sizing plans for 10.6 GB minimum memory and 17.7 GB recommended memory, with the Q4_K_M checkpoint requiring roughly 10.2 GB of VRAM and 9.3 GB of SSD storage. Exceeding dedicated VRAM forces PCIe weight swapping across host RAM and slows step sampling dramatically.
Desktop NVIDIA RTX workstations using ComfyUI or Diffusers with CUDA acceleration deliver the cleanest generation throughput, while high-memory Apple Silicon Macs can execute workloads via MPS with conservative batch size 1 and lower output dimensions. Under the LTX-2 Community license, verify commercial deployment terms on the official repository card before production use.
Highlights
- Generation VRAM Budget10.6 GB minimum and 17.7 GB recommended across backbone, encoder, and VAE buffers.
- Baseline QuantizationQ4_K_M targets 10.2 GB memory and 9.3 GB storage.
- Recommended PipelineComfyUI or Diffusers on NVIDIA CUDA desktop GPUs or high-RAM Apple Silicon Macs.
System requirements by device
Memory planning targets 10.6 GB minimum RAM/VRAM and 17.7 GB recommended memory. The baseline quant is Q4_K_M, estimated at 10.2 GB memory and 9.3 GB disk. Actual runtime usage shifts with context length, KV cache precision, and OS background overhead.
| Device | Notes |
|---|---|
| Mac (Apple silicon) | Use unified memory headroom (~75% physical RAM max) as the primary fit signal. Start with Q4_K_M and close heavy browser tabs first. |
| Windows / Linux GPU | Prefer dedicated NVIDIA RTX VRAM via CUDA (ComfyUI / Diffusers) so UNet, text encoder, and VAE buffers stay off PCIe RAM. |
| Phones & Compact Edge Boards | Not supported locally: mobile RAM ceilings and app watchdogs cannot hold multi-stage diffusion and VAE buffers. |
Start with a short session and leave memory for the operating system. A file fitting on disk does not guarantee that it will load in memory.
Estimated memory by quantization
These are rough estimates, not tested minimum requirements. Check the actual file and runtime documentation; context, offloading and operating-system overhead can change the memory needed.
| Quant Tier | Estimated memory | Disk Size | Quality |
|---|---|---|---|
| Q2_K | 6.6 GB | 5.8 GB | low |
| Q3_K_M | 9.0 GB | 8.1 GB | moderate |
| Q4_K_M | 10.2 GB | 9.3 GB | good |
| Q5_K_M | 12.7 GB | 11.6 GB | good |
| Q6_K | 15.1 GB | 13.9 GB | excellent |
| Q8_0 | 20.0 GB | 18.6 GB | excellent |
| F16 | 39.4 GB | 37.2 GB | lossless |
How to run LTX 2.3 on your device
Start with a short prompt and the smallest quant that still gives the quality you need. A clean first run is more useful than loading a larger file with no memory headroom.
Run LTX 2.3 on Windows / Linux GPU
- Install ComfyUI or a current Diffusers Python environment with PyTorch CUDA enabled.
- Select a quantized checkpoint such as Q4_K_M (10.2 GB) and confirm at least 10.6 GB of free VRAM.
- Keep VAE tiling enabled and test a standard resolution (e.g., 1024x1024 or short frame count) before increasing batch or step counts.
- Avoid spilling diffusion UNet/DiT weights across PCIe into system RAM if interactive generation speed matters.
Run LTX 2.3 on Apple Silicon Mac
- Confirm your Mac has at least 17.7 GB unified memory so Metal buffers do not force macOS swap.
- Load Q4_K_M weights inside ComfyUI (MPS) or a native Apple Silicon Diffusers workflow.
- Close memory-heavy browsers and keep latent batch size at 1 during initial generation tests.
- Consult the Mac hardware guide when planning multi-step image or video pipelines.
LTX 2.3 vs alternatives
| Model | Params | Context | Min RAM | License |
|---|---|---|---|---|
| LTX 2.3 | 19B | 4K | 10.6 GB | LTX-2 Community |
| Qwen Image 2512 | 20B | 4K | 11.2 GB | Apache 2.0 |
| GPT-OSS 20B | 21B (3.6B active) | 128K | 11.7 GB | Apache 2.0 |
| GLM-Image | 16B (16B) | 4K | 11 GB | MIT |
| Qwen 2.5 14B | 14B | 128K | 7.8 GB | Apache 2.0 |
| Phi-4 14B | 14B | 16K | 7.8 GB | MIT |
Use this table to choose the smallest nearby model that still matches the job. A related model with lower RAM can be the better local choice when the current page fits only at the edge of available memory.
Frequently asked questions
- What is LTX 2.3?
- LTX 2.3 is a 19B media generation model from Lightricks. Sizing estimates allocate 10.6 GB minimum VRAM and 17.7 GB recommended memory for weights, text encoders, and VAE decode buffers.
- How do I download and install ltx 2.3?
- Download a GGUF build for LTX 2.3, start with Q4_K_M, and confirm the model fits locally before increasing context or quant quality.
- How much VRAM or RAM does LTX 2.3 require?
- At Q4_K_M, LTX 2.3 occupies roughly 10.2 GB of active memory and 9.3 GB on disk. Add 2–6 GB of spare headroom for KV cache buffers, runtime scratch space, and OS apps (17.7 GB recommended).
- Which software runs LTX 2.3 locally?
- Run LTX 2.3 inside ComfyUI or Hugging Face Diffusers on an NVIDIA CUDA desktop GPU or high-memory Apple Silicon Mac. Start with Q4_K_M or FP8 builds before testing higher-precision weights.
- Can phones run LTX 2.3 locally?
- No. Desktop image and video diffusion checkpoints require high-bandwidth desktop GPUs or high-memory Apple Silicon Macs rather than smartphone runtimes.
- How does LTX 2.3 compare to Qwen Image 2512?
- Qwen Image 2512 (20B, 11.2 GB min RAM) sits closest in size. Compare active parameters, context ceiling (4K vs 4K), and license terms (LTX-2 Community) before switching.