LTX-2.3 Distilled FP8
LTX-2.3 is a 22B video model from Lightricks with synchronized audio generation. The catalog offers distilled and FP8 configurations. Resolution, frame rate and generation time depend on the selected workflow and GPU.
- GPU tier
- GPU High VRAM (A6000)
- Estimated customer rate
- $0.72/hr
- VRAM
- 48GB
- Template
- comfyui-ltx2
LTX-2.3 Distilled
LTX-2.3 is a 22B video model from Lightricks with synchronized audio generation. The catalog offers distilled and FP8 configurations. Resolution, frame rate and generation time depend on the selected workflow and GPU.
- GPU tier
- GPU Very High VRAM (A100 80GB)
- Estimated customer rate
- $1.85/hr
- VRAM
- 80GB
- Template
- comfyui-ltx2
Deployment facts
| Factor | LTX-2.3 Distilled FP8 | LTX-2.3 Distilled |
|---|---|---|
| Model family | ltx23 | ltx23 |
| Variant | Distilled FP8 (24GB GPU) | Distilled (Recommended) |
| Type | video | video |
| Deployment template | comfyui-ltx2 | comfyui-ltx2 |
| GPU tier | GPU High VRAM (A6000) | GPU Very High VRAM (A100 80GB) |
| GPU | NVIDIA RTX A6000 | NVIDIA A100 80GB PCIe |
| VRAM | 48GB | 80GB |
| vCPU / RAM / disk | 9 vCPU / 50GB RAM / 200GB disk | 8 vCPU / 117GB RAM / 300GB disk |
| Estimated customer rate | $0.72/hr | $1.85/hr |
| Estimated 730-hour month | $522/mo | $1348/mo |
| Paid launch | Available; review setup before launch | Available; review setup before launch |
Estimates use the same rate as the model catalog, including configured disk and the ModelPilot service fee. GPU time is billed during startup and while allocated. Configuration changes and retained storage can affect your total; review the launch details before paying.
Cost scenarios
| Usage level | Hours/mo | LTX-2.3 Distilled FP8 | LTX-2.3 Distilled |
|---|---|---|---|
| Prototype | 40 | $28.60 | $73.84 |
| Part-time app | 160 | $114 | $295 |
| Always-on | 730 | $522 | $1348 |
What this comparison covers
- LTX-2.3 Distilled FP8: 48GB GPU memory, estimated $0.72/hr including configured disk and the service fee.
- LTX-2.3 Distilled: 80GB GPU memory, estimated $1.85/hr including configured disk and the service fee.
- Review each template’s required input files, model downloads and manual setup steps before launching.
- Dedicated GPU time is billed during startup and while allocated. Retained storage can remain billable after stopping.
- These are deployment comparisons, not output-quality benchmarks. Test representative inputs and review the results for your project.
Configuration differences
| Factor | LTX-2.3 Distilled FP8 | LTX-2.3 Distilled | Lower cost or requirement |
|---|---|---|---|
| Lower estimated customer rate | $0.72/hr | $1.85/hr | LTX-2.3 Distilled FP8 |
| Lower VRAM requirement | 48GB | 80GB | LTX-2.3 Distilled FP8 |
| Cheaper always-on deployment | $522/mo | $1348/mo | LTX-2.3 Distilled FP8 |
FAQ
Which is cheaper to run, LTX-2.3 Distilled FP8 or LTX-2.3 Distilled?
LTX-2.3 Distilled FP8 has the lower estimated customer rate at $0.72/hr, including configured disk and the ModelPilot service fee. Startup is billed; retained storage and configuration changes can affect your total.
Which uses less VRAM, LTX-2.3 Distilled FP8 or LTX-2.3 Distilled?
LTX-2.3 Distilled FP8 maps to the lower VRAM tier at 48GB in the current deployment recommendation.
Can both models run through ModelPilot?
Both have a deployment option in the catalog. Review each model's setup requirements, available GPU and estimated rate before launching. Output quality depends on the model, input and workflow.