Bonus: Top up now and we'll double your first deposit — get x2 credits instantly.

LTX-2.5: Self-Hosting & Deployment Guide

LTX-2.5 is an open-weight image-to-video generation model.

What is LTX-2.5?

LTX-2.5 is an open-weight image-to-video generation model, published by Lightricks with open weights at Lightricks/LTX-2.5. LLM API does not route this model today, so there is no endpoint or code sample for it here. What this page does give you is the self-hosting picture: published weight sizes, the VRAM each quantisation needs and the commands to serve it yourself. Every figure is read from the public repository and refreshed weekly. At import it had 1,638,605 downloads on Hugging Face in the previous 30 days, under the other licence.

How to run LTX-2.5 yourself

Open weights (other). Below: how much memory the weights need at each quantization, which hardware fits them, and the commands to serve it.

LTX-2.5 VRAM requirements

Precision / quantizationWeightsFigureSource
Full precision weights200.9 GBFile sizeLightricks/LTX-2.5
FP880.2 GBFile sizeChrisColeTech/LTX-2.5-uncensored-v1.1-FP8
MLX 4-bit38.8 GBFile sizeddalcu/LTX-2.5-MLX-Serve-4bit
GGUF Q8_023.6 GBFile sizeAbiray/LTX-2.5-Distilled-GGUF
GGUF Q4_K_M15.7 GBFile sizeAbiray/LTX-2.5-Distilled-GGUF

File size = actual download size on Hugging Face. Estimate = parameter count × bytes per weight. Weights only — the KV cache for long contexts needs extra memory; the fit tables below add 15% headroom as a rule of thumb.

GPU requirements: deploy LTX-2.5 on your hardware

HardwareFull precision weightsFP8MLX 4-bitGGUF Q8_0GGUF Q4_K_M
NVIDIA H200141 GB2×1×1×1×1×
NVIDIA H10080 GB3×2×1×1×1×
NVIDIA A100 80GB80 GB3×2×1×1×1×
NVIDIA L40S48 GB5×2×1×1×1×

GPUs needed per variant.

HardwareGGUF Q8_0GGUF Q4_K_M
RTX 509032 GB1×1×
RTX 409024 GB2×1×
RTX 309024 GB2×1×

GGUF via llama.cpp; more cards or CPU offload for larger files.

HardwareGGUF Q8_0GGUF Q4_K_MMLX 4-bit
Mac, 512 GBM3 UltraFitsFitsFits
Mac, 192 GBM2 UltraFitsFitsFits
Mac, 128 GBM4 Max / M3 MaxFitsFitsFits
Mac, 64 GBM4 Pro / MaxFitsFitsFits

Assumes ~75% of unified memory is usable by the GPU.

HardwareFull precision weightsFP8MLX 4-bitGGUF Q8_0GGUF Q4_K_M
AWS p5.48xlarge8× H100 · 640 GBFitsFitsFitsFitsFits
AWS p4de.24xlarge8× A100 80GB · 640 GBFitsFitsFitsFitsFits
AWS g6e.12xlarge4× L40S · 192 GB—FitsFitsFitsFits
Google Cloud a3-highgpu-8g8× H100 · 640 GBFitsFitsFitsFitsFits
Azure ND H100 v58× H100 · 640 GBFitsFitsFitsFitsFits

Whole-instance GPU memory; public instance specs.

Not on LLM.API yet

We do not route this model through our API at the moment. Browse the models you can call today — most workloads have a close match already live.

Not available on LLM API yet

We do not route this model through our API at the moment, so there is no endpoint or code snippet for it yet. Browse the models you can call today — most workloads have a close match already live.

Browse available models

When to Use — When NOT to Use

Use it if...

  • You need images generated from a text prompt, or edited from a reference image (LTX-2.5)
  • You want image generation on the same LLM API key as your text models, with no extra vendor SDK
  • Your pipeline can take the result as a URL or base64 payload rather than a streamed response
  • You want per-image pricing you can read off the catalogue before you commit

Avoid if...

  • You need a chat, coding or reasoning model — this model returns images, not text
  • You need token-by-token streaming: image calls return once the picture is finished
  • You need video, speech or embeddings; those are separate models in the catalogue
  • Your licensing review has not cleared generated imagery for commercial use

Get one key to every model

Swap your API key. Keep your code.