Ornith-1.0-397B GPU Requirements: VRAM & Cheapest GPU
Ornith-1.0-397B has about 397B parameters. See exactly how much GPU memory it needs at FP16, INT8, and INT4, and the cheapest GPU to run it, with live hourly pricing from 5+ data center partners.
- Parameters
- 397B
- Min VRAM (INT4)
- 216 GB
- Cheapest (INT4)
- $4.76/hr
- Deploy
- < 2 min
To run Ornith-1.0-397B for inference at FP16, you need roughly 865 GB of VRAM. The cheapest fit on Spheron AI is 4x B300 288GB at about $29.88/hr. Quantize to INT4 to run it on a smaller, cheaper GPU.
VRAM required
Estimated peak VRAM including weights, activations, and KV cache. Add 10% headroom for production traffic.
- 01$7.47/hr$29.88/hr4× B300 288GBCHEAPESTBlackwell Ultra · HBM3e
- 02$4.80/hr$38.40/hr8× H200 141GBHopper · HBM3e
- 03$11.06/hr$88.48/hr8× B200 192GBBlackwell · HBM3e
Live pricing aggregated from 5+ data center partners. Per-minute billing, no commitments.
VRAM required to run Ornith-1.0-397B
Estimated peak VRAM at context length 4,096 and batch size 1, including weights, activations, and KV cache. Quantizing to INT8 (Q8) or INT4 (Q4) cuts memory roughly in half and in quarter.
| Precision | Inference | LoRA fine-tune | Full fine-tune |
|---|---|---|---|
| FP16 | 865 GB | 1298 GB | 3460 GB |
| INT8 | 433 GB | 649 GB | 1730 GB |
| INT4 | 216 GB | 324 GB | 865 GB |
Cheapest GPU to run Ornith-1.0-397B by precision
865GB
Full precision. Best quality, highest memory.
433GB
8-bit quantized. ~2x smaller, minimal quality loss.
216GB
4-bit quantized. ~4x smaller, runs on smaller GPUs.
Inference vs fine-tuning Ornith-1.0-397B
Inference only holds the model weights plus a KV cache, so it is the cheapest setup. LoRA fine-tuning adds a small adapter and roughly 50% more memory. Full fine-tuning holds gradients and optimizer state on top of the weights, which is about 4x the inference footprint, so it often needs multiple GPUs even when inference fits on one. For Ornith-1.0-397B, an on-demand B300 288GB instance covers inference and LoRA, while a full fine-tune needs several times that memory and often spans multiple GPUs. Check the live GPU pricing for current rates.
Similar models
Compare GPU requirements for models in the same class.
Ornith-1.0-397B GPU questions
VRAM, GPU choice and cost questions for running Ornith-1.0-397B on cloud GPUs.
Ornith-1.0-397B has about 397B parameters. At FP16 it needs roughly 865 GB of VRAM for inference, including weights, activations, and KV cache. Quantized to INT4 that drops to around 216 GB. Leave about 10% headroom for production traffic.
For FP16 inference, the cheapest fit on Spheron AI is 4x B300 288GB at about $29.88/hr. If you quantize to INT4, 4x A100 80GB from $4.76/hr runs it for less. Pricing is aggregated live from 5+ data center partners with per-minute billing.
LoRA fine-tuning adds roughly 50% on top of inference memory, so it usually fits the same class of GPU. Full fine-tuning holds gradients and optimizer state and needs about 4x the inference VRAM, which often means multiple GPUs for Ornith-1.0-397B. The VRAM matrix above shows the exact estimate for each setup.
We read the parameter count directly from the model's safetensors metadata on HuggingFace, then estimate peak VRAM from weights, activations, KV cache, and framework overhead at your chosen precision. The estimate lands within about 15% of real-world use for most transformer models.
Quantized to INT4, Ornith-1.0-397B needs about 216 GB of VRAM, so it is too large for a single 24 GB RTX 4090 and needs a bigger card or multiple GPUs. At FP16 it needs roughly 865 GB, which usually means a data center GPU. The precision picks above list the cheapest GPU that fits each setup.
No. Even quantized to INT4, Ornith-1.0-397B needs about 216 GB of VRAM, so it takes 3 x 80 GB GPUs at minimum, and 11 x 80 GB GPUs at FP16. The precision picks above show the cheapest GPU that fits each setup on Spheron AI.
Ornith-1.0-397B needs 11 x 80 GB GPUs at FP16 (about 865 GB of VRAM), 6 x 80 GB GPUs at INT8 (433 GB) and 3 x 80 GB GPUs at INT4 (216 GB). At FP16 that is more than one 8-GPU node, so plan for a multi-node cluster with a fast interconnect. Smaller cards such as 48 GB or 24 GB GPUs need proportionally more of them.
Ornith-1.0-397B needs about 433 GB of VRAM at INT8, roughly half its FP16 footprint of 865 GB. The cheapest INT8 fit on Spheron AI is 8x A100 80GB from $9.52/hr. INT8 is the usual middle ground: half the memory of FP16, with a smaller quality drop than 4-bit quantization.
Running Ornith-1.0-397B around the clock at FP16 costs about $21,812 a month on 4x B300 288GB (730 hours at $29.88/hr). Quantized to INT4 it costs about $3,475 a month on 4x A100 80GB (730 hours at $4.76/hr at the spot rate, which can be reclaimed). Spheron AI bills per minute, so stopping the instance when it is idle cuts that figure directly.
Ornith-1.0-397B needs about 865 GB at FP16, 433 GB at INT8 and 216 GB at INT4. Use FP16 when output quality matters most or you plan to fine-tune, INT8 for most production inference, and INT4 when cost or GPU size is the constraint: the cheapest INT4 fit is 4x A100 80GB from $4.76/hr.