What GPU do I need to run Nanbeige/Nanbeige4.3-33B?
33.1B parameters, published in BF16. View on Hugging Face
Nanbeige4.3-33B is published by Nanbeige on Hugging Face, with 20,702 downloads and 1,069 likes to date. It's a LlamaForCausalLM 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 | 61.7 GB | 74.0 GB | A100 | 1 | $1.06/hr |
| cheaper alt. | RTX A4000 | 5 | $0.684/hr | ||
| FP8 (quantized) | 30.8 GB | 37.0 GB | L40 | 1 | $0.759/hr |
| cheaper alt. | RTX 4000 Ada | 2 | $0.440/hr | ||
| INT4 (quantized) | 15.4 GB | 18.5 GB | RTX A5000 | 1 | $0.176/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 Nanbeige4.3-33B at its published (BF16) precision: 1× A100, at $1.06/hr per GPU ($1.06/hr total). Quantizing to FP8 or INT4 (rows above) can cost less, but requires a compatible quantized checkpoint to exist for this model.
Nanbeige4.3-33B: 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.
Grouped-query attention: every layer caches keys and values for a few shared KV heads. Cached per token: 212,992 bytes at 16-bit.
| Context | KV cache, one sequence |
|---|---|
| 4K | 0.81 GB |
| 32K | 6.5 GB |
| 128K | 26.0 GB |
| 512K (model maximum) | 104.0 GB |
Computed from the layer, head and window counts in the model's own configuration (Nanbeige/Nanbeige4.3-33B), at 16 bits per cached value. Each concurrent sequence needs its own cache, on top of the weights.
Nanbeige4.3-33B: common questions
Can Nanbeige4.3-33B run on a single GPU?
Yes, but not on a desktop card. At BF16 it needs 74.0 GB of VRAM, more than a single 32 GB desktop card holds. The cheapest live single-card fit is a 80.0 GB A100 at $1.06/hr.
What is the least VRAM Nanbeige4.3-33B can run in?
18.5 GB, at INT4 (quantized), which fits a 24 GB card, against 74.0 GB at BF16. That row assumes an INT4 checkpoint has actually been published for this model, so check its Hugging Face page before planning around it.
Does quantizing Nanbeige4.3-33B lower the GPU bill?
Yes. At BF16 the cheapest live fit is one A100 at $1.06/hr. At INT4 (quantized) it drops to one RTX A5000 at $0.176/hr, provided a quantized checkpoint exists for it.
Weight-to-VRAM math, the fit rules, and how live prices are normalized: full methodology.
More Nanbeige models
- Nanbeige4.1-3B (3.9B, BF16)
- Nanbeige4.2-3B (4.2B, BF16)
- Nanbeige4.2-3B-DSpark (848M, BF16)
Alternatives at this size
Other models for text-generation within about a third of Nanbeige4.3-33B's 33.1B parameters, from other model lines.
- Qwen3-32B (32.8B, BF16)
- GLM-4.7-Flash (31.2B, BF16)
- NVIDIA-Nemotron-3-Nano-30B-A3B-BF16 (31.6B, BF16)
- granite-4.1-30b (28.9B, BF16)
- Ornith-1.5-35B-A3B (36.0B, BF16)
More on Nanbeige4.3-33B
Fits on an 80 GB GPU at BF16: every model that fits in 80 GB.
Fits on a 48 GB GPU at FP8: every model that fits in 48 GB.
Fits on a 24 GB GPU at INT4: every model that fits in 24 GB.
Best chat and assistants models: how Nanbeige4.3-33B ranks against the rest.
Related reading: A100 pricing and specs, How much VRAM you need for LLMs, Serving LLMs with vLLM, vLLM vs TensorRT-LLM vs SGLang, and Best GPU for LLM inference.
Your own GPU cloud accounts
One console for the GPU clouds you already rent from. Your keys stay on your machine.
$20 a month, flat