Lightricks · Video · 22B parameters · other
LTX-2.5
Needs 27 GB at FP8 / INT8. The smallest real configuration today is 1× Tesla V100 32GB at Vast.ai, $0.07 per hour, about $50 a month at 730 hours.
Parameters22Bdense
CategoryVideosized on weights + headroom
Checkpoint74 GBBF16, as published
Licenseothercheck before commercial use
ReleasedJul 23, 2026documented date
Smallest configuration · FP8 / INT8
1× Tesla V100 32GB32 GB VRAM · 83% used
- Per hour
- $0.07
- Per month
- $50
- Cheapest at
Vast.ai
What is documented about LTX-2.5
news checked 2026-10-27Some listings give 19B; the published weight files are named 22b, and the repo is 70 GB in BF16 (read 2026-10-27).
- https://huggingface.co/Lightricks/LTX-2.5
- https://huggingface.co/api/models/Lightricks/LTX-2.5?blobs=true
Memory · 3 precisionsHow much VRAM it needs
Weights plus 20% runtime overhead. For video models, activations also depend on resolution, clip length and batch size; the per-GPU estimates below give the practical configuration.
| Precision | Weights | Weights + headroom |
|---|---|---|
| BF16published checkpoint | 74 GB | 89 GB |
| FP8 / INT8estimated | 22 GB | 27 GB |
| INT4 / MXFP4estimated | 13 GB | 16 GB |
Cross-checked on Hugging Face Oct 28, 2026: repository read, 73.8 GB of weight files. Method: weights as published at the native precision, verified on Hugging Face 2026-10-28; other precisions estimated at 1 (8-bit) and 0.6 (4-bit) bytes per parameter; × 1.2 runtime headroom; KV cache not added (architecture not readable).
Smallest configuration per GPU · 110 GPUs · cheapest eligible on-demand offer per GPU count, every provider on fileWhat it costs to run, GPU by GPU
The fewest GPUs of one type whose total memory fits the model at short context, priced from the cheapest eligible on-demand offer for exactly that count (45-day window, newest reading per provider), with the next providers beside it. Monthly = 730 hours.
| GPU | Smallest config | VRAM used | $ / hour | $ / month | Cheapest at |
|---|---|---|---|---|---|
| Tesla V100 32GBBest fit | 1× 32 GB32 GB total | 83% | $0.07 | $50 | Vast.aiTrust 78 · read Nov 2 · receipt · also Cyfuture AI $0.57, E2E Networks $1.50 |
| T4 | 2× 16 GB32 GB total | 83% | $0.08 | $59 | Vast.aiTrust 78 · read Nov 2 · receipt · also iwinv $0.39, Theta EdgeCloud $0.40 |
| Tesla P100 | 2× 16 GB32 GB total | 83% | $0.10 | $70 | Vast.aiTrust 78 · read Nov 2 · receipt |
| RTX A4000 | 2× 16 GB32 GB total | 83% | $0.14 | $105 | Vast.aiTrust 78 · read Nov 2 · receipt · also Clore.ai $0.19, AtmosCompute $0.30 |
| RTX PRO 2000 | 2× 16 GB32 GB total | 83% | $0.15 | $106 |  Clore.aiTrust 46 · read Nov 2 · receipt |
| RTX 3060 | 4× 12 GB48 GB total | 55% | $0.16 | $118 | Clore.aiTrust 46 · read Nov 2 · receipt · also Vast.ai $0.21, Edgevana $0.31 |
| RTX 5060 Ti | 2× 16 GB32 GB total | 83% | $0.19 | $137 | Vast.aiTrust 78 · read Nov 2 · receipt · also Clore.ai $0.19, Edgevana $0.31 |
| Tesla V100 16GB | 2× 16 GB32 GB total | 83% | $0.19 | $138 | Vast.aiTrust 78 · read Nov 2 · receipt · also CloudRift $0.50, AtmosCompute $0.65 |
| Quadro RTX 5000 | 2× 16 GB32 GB total | 83% | $0.20 | $148 | Vast.aiTrust 78 · read Nov 2 · receipt · also OVHcloud $1.21 |
| GTX 1080 | 4× 8 GB32 GB total | 83% | $0.20 | $150 | Vast.aiTrust 78 · read Nov 2 · receipt |
| RTX 3090 | 2× 24 GB48 GB total | 55% | $0.21 | $157 | Vast.aiTrust 78 · read Nov 2 · receipt · also Clore.ai $0.24, Lium $0.44 |
| RTX 2070 | 4× 8 GB32 GB total | 83% | $0.22 | $157 | Vast.aiTrust 78 · read Nov 2 · receipt · also Edgevana $0.25 |
| RTX 5090 | 1× 32 GB32 GB total | 83% | $0.22 | $158 | Clore.aiTrust 46 · read Nov 2 · receipt · also Salad $0.24, HyperAI $0.35 |
| RTX 3060 Ti | 4× 8 GB32 GB total | 83% | $0.22 | $159 | Vast.aiTrust 78 · read Nov 2 · receipt · also Clore.ai $0.29 |
| Quadro RTX 8000 | 1× 48 GB48 GB total | 55% | $0.23 | $167 | Vast.aiTrust 78 · read Nov 2 · receipt · also TensorDock $0.32 |
| GTX 1070 Ti | 4× 8 GB32 GB total | 83% | $0.23 | $169 | Vast.aiTrust 78 · read Nov 2 · receipt |
| RTX 2070 Super | 4× 8 GB32 GB total | 83% | $0.24 | $177 | Vast.aiTrust 78 · read Nov 2 · receipt · also Edgevana $0.28 |
| Titan RTX | 2× 24 GB48 GB total | 55% | $0.27 | $196 | Vast.aiTrust 78 · read Nov 2 · receipt · also Edgevana $0.52 |
| Titan Xp | 4× 12 GB48 GB total | 55% | $0.27 | $196 | Vast.aiTrust 78 · read Nov 2 · receipt |
| L40 | 1× 48 GB48 GB total | 55% | $0.27 | $196 | Vast.aiTrust 78 · read Nov 2 · receipt · also io.net $0.66, Runpod $0.69 |
| RTX 2080 Ti | 4× 11 GB44 GB total | 60% | $0.28 | $202 | Vast.aiTrust 78 · read Nov 2 · receipt |
| RTX 4070 Ti Super | 2× 16 GB32 GB total | 83% | $0.29 | $213 | Clore.aiTrust 46 · read Nov 2 · receipt · also AtmosCompute $0.58, Edgevana $0.77 |
| RTX 5070 Ti | 2× 16 GB32 GB total | 83% | $0.32 | $235 | Vast.aiTrust 78 · read Nov 2 · receipt · also Clore.ai $0.36, Edgevana $0.46 |
| RTX A6000 | 1× 48 GB48 GB total | 55% | $0.33 | $241 |  GPURentonot scored · read Nov 2 · receipt · also gcube $0.34, Thunder Compute $0.35 |
| The 24 cheapest of 82 GPU types that fit. Every GPU, any precision and context, in the finder | |||||
Throughput estimates · Oct 4, 2026Estimated speed, in min per clip
Video. Speed estimated from memory bandwidth and model size, never measured. The cost per video second is an estimate at the prices read on Oct 4, 2026; it is not recomputed here.
| GPU | Cards | Precision | min per clip | per video second · estimate |
|---|---|---|---|---|
| RTX 5090 | 1 | FP8 | 5.5 | $0.009 |
| RTX 4090 | 2 | FP8 | 5.8 | $0.009 |
| L40S | 1 | FP8 | 7.6 | $0.014 |
| RTX PRO 6000 | 1 | FP8 | 3.9 | $0.015 |
| H100 SXM | 1 | FP8 | 3.1 | $0.019 |
| H100 PCIe | 1 | FP8 | 4.3 | $0.019 |
| B200 | 1 | FP8 | 1.6 | $0.021 |
| A100 80GB | 1 | BF16 | 7.3 | $0.023 |
| H200 | 1 | FP8 | 2.9 | $0.024 |
| MI300X | 1 | FP8 | 4.5 | $0.024 |
cards, precision and speed estimated from memory bandwidth and model size; cost recomputed by the site from verified prices · cheapest per output at the prices of Oct 4, 2026: RTX 5090Why estimates, and their limits →
LTX-2.5 · related