What GPU do I need to run ricursiveAI/Ripple-65B-Preview?
64.8B parameters, published in BF16. View on Hugging Face
Set up Ripple-65B-PreviewRipple-65B-Preview is published by ricursiveAI on Hugging Face, with 5,835 downloads and 438 likes to date. It's a RicursiveForCausalLM model built for text-generation, published natively in BF16.
VRAM required & cheapest live GPU fit
Required VRAM = weight size at each precision times a flat 1.2 overhead for activations and fragmentation. The KV-cache grows with context and is not in that factor; it is listed per model below where the architecture is published. Full formula and assumptions: methodology.
| Precision | Weight size | Required VRAM | Cheapest live fit | GPUs needed | Est. $/hr (full fit) |
|---|---|---|---|---|---|
| BF16 | 120.7 GB | 144.8 GB | RTX 3090 | 7 | $1.13/hr |
| FP8 (quantized) | 60.3 GB | 72.4 GB | RTX PRO 6000 | 1 | $1.38/hr |
| cheaper alt. | RTX 5060 Ti | 5 | $0.825/hr | ||
| INT4 (quantized) | 30.2 GB | 36.2 GB | A100 | 1 | $0.462/hr |
| cheaper alt. | RTX 3090 | 2 | $0.323/hr |
A GPU is only matched to a row if its hardware supports that precision, and the primary recommendation is always a single-GPU fit when one exists.
INT4 requires a quantized checkpoint actually published for this model, check its Hugging Face page before relying on this row.
Cheapest way to run Ripple-65B-Preview at its published (BF16) precision: 7× RTX 3090, at $0.162/hr per GPU ($1.13/hr total). Quantizing to FP8 or INT4 (rows above) can cost less, but requires a compatible quantized checkpoint to exist for this model.
Ripple-65B-Preview: KV cache by context length
The KV cache is the memory the attention layers hold for every token of context, on top of the weights. It is not part of the flat 1.2x overhead in the table above, grows with context length and with every concurrent request, and is why a long-context deployment needs more VRAM than the table shows.
Sliding-window attention: windowed layers stop growing at the window, the full-attention layers keep growing. Cached per token: 327,680 bytes at 16-bit while the context is inside the window.
| Context | KV cache, one sequence |
|---|---|
| 4K | 0.50 GB |
| 32K | 2.3 GB |
| 128K | 8.3 GB |
| 2M (model maximum) | 128.3 GB |
Computed from the layer, head and window counts in the model's own configuration (ricursiveAI/Ripple-65B-Preview), at 16 bits per cached value. Each concurrent sequence needs its own cache, on top of the weights.
Ripple-65B-Preview: common questions
Can Ripple-65B-Preview run on a single GPU?
No. At BF16 it needs 144.8 GB of VRAM, more than a single 32 GB desktop card holds. The cheapest capable card in the live feed is a 24.0 GB RTX 3090, and it takes 7 of them.
How many GPUs do I need to run Ripple-65B-Preview?
7 at BF16. It needs 144.8 GB of VRAM and the cheapest capable live offer is a 24.0 GB RTX 3090, so 7 of them come to $1.13/hr in total.
Does quantizing Ripple-65B-Preview lower the GPU bill?
Yes. At BF16 the cheapest live fit is 7 RTX 3090 cards at $1.13/hr. At INT4 (quantized) it drops to one A100 at $0.462/hr, provided a quantized checkpoint exists for it.
Weight-to-VRAM math, the fit rules, and how live prices are normalized: full methodology.
More ricursiveAI models
- Ripple-65B-Base-Preview (64.8B, BF16)
- Ripple-65B-Preview-DFlash2 (4.0B, BF16)
Lineage
Quantized and fine-tuned versions
More on Ripple-65B-Preview
Fits on a 192 GB GPU at BF16: every model that fits in 192 GB.
Fits on an 80 GB GPU at FP8: every model that fits in 80 GB.
Fits on a 48 GB GPU at INT4: every model that fits in 48 GB.
Related reading: RTX 3090 pricing and specs, How much VRAM you need for LLMs, and What AI inference is.
