What GPU do I need to run meta-llama/Llama-3.3-70B-Instruct?
Meta's 70B multilingual instruction-tuned chat model. 70.6B parameters, published in BF16. View on Hugging FaceGated
Llama-3.3-70B-Instruct is published by meta-llama on Hugging Face, with 549,302 downloads and 2,989 likes to date. It's a LlamaForCausalLM model built for text-generation, published natively in BF16, and gated: you'll need to accept the model's terms on Hugging Face before downloading weights.
What Llama 3.3 70B Instruct is
Llama 3.3 70B Instruct is Meta's instruction-tuned chat model, built for assistant-style dialogue in eight languages (English, German, French, Italian, Portuguese, Hindi, Spanish and Thai). Meta's card positions it for commercial and research use, including as a source model for synthetic data generation and distillation into smaller models.
License note: Meta's own commercial license, not OSI-approved open source; the repo is gated on Hugging Face, so you accept its terms there before downloading weights. Facts in this section are sourced from Llama 3.3 70B Instruct's Hugging Face model card, not benchmarked by Aquanode.
What it's used for
- Multilingual chat assistants
- Instruction following
- Synthetic data generation
- Distillation into smaller models
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 | 131.4 GB | 157.7 GB | RTX A5000 | 7 | $1.23/hr |
| FP8 (quantized) | 65.7 GB | 78.8 GB | RTX PRO 6000 | 1 | $1.51/hr |
| cheaper alt. | RTX 5060 Ti | 5 | $0.550/hr | ||
| INT4 (quantized) | 32.9 GB | 39.4 GB | RTX A6000 | 1 | $0.363/hr |
| cheaper alt. | RTX 5060 Ti | 3 | $0.330/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 Llama-3.3-70B-Instruct at its published (BF16) precision: 7× RTX A5000, at $0.176/hr per GPU ($1.23/hr total). Quantizing to FP8 or INT4 (rows above) can cost less, but requires a compatible quantized checkpoint to exist for this model.
Llama-3.3-70B-Instruct: 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: 327,680 bytes at 16-bit.
| Context | KV cache, one sequence |
|---|---|
| 4K | 1.3 GB |
| 32K | 10.0 GB |
| 128K | 40.0 GB |
Computed from the layer, head and window counts in the model's own configuration (unsloth/Llama-3.3-70B-Instruct), at 16 bits per cached value. Each concurrent sequence needs its own cache, on top of the weights.
Llama-3.3-70B-Instruct: common questions
Can Llama-3.3-70B-Instruct run on a single GPU?
No. At BF16 it needs 157.7 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 A5000, and it takes 7 of them.
Do I need approval to download Llama-3.3-70B-Instruct?
Yes. meta-llama gates this repository on Hugging Face, so you have to accept its terms with a signed-in account before the weights will download. It does not change the 157.7 GB the model needs once you have them.
How many GPUs do I need to run Llama-3.3-70B-Instruct?
7 at BF16. It needs 157.7 GB of VRAM and the cheapest capable live offer is a 24.0 GB RTX A5000, so 7 of them come to $1.23/hr in total.
Does quantizing Llama-3.3-70B-Instruct lower the GPU bill?
Yes. At BF16 the cheapest live fit is 7 RTX A5000 cards at $1.23/hr. At INT4 (quantized) it drops to one RTX A6000 at $0.363/hr, provided a quantized checkpoint exists for it.
How to run Llama 3.3 70B Instruct
Run Llama 3.3 70B Instruct with vLLM
Generic example, not from the model's own docs: adjust flags (quantization, context length, parallelism) for your setup.
vllm serve meta-llama/Llama-3.3-70B-Instruct --tensor-parallel-size 7Run Llama 3.3 70B Instruct with Ollama
Verified against Ollama's own library listing.
ollama run llama3.3:70bRun Llama 3.3 70B Instruct with GGUF quantizations
Prebuilt GGUF weights published at unsloth/Llama-3.3-70B-Instruct-GGUF. Run with llama.cpp's llama-server or load the repo directly in LM Studio.
llama-server -hf unsloth/Llama-3.3-70B-Instruct-GGUFSource: https://huggingface.co/unsloth/Llama-3.3-70B-Instruct-GGUF
Deploy Llama 3.3 70B Instruct on Aquanode
Aquanode has no one-click deploy template for Llama 3.3 70B Instruct; you install the inference engine yourself with the commands below. Aquanode sells GPU pods billed per second, not a hosted inference API.
- Launch a bare GPU pod sized to the requirement above (7× RTX A5000 or larger).
- Open a terminal on the pod, or save one of the commands above as a startup script so it runs automatically the first time the pod boots.
- Run the command and connect to the resulting endpoint.
Launch a GPU podWeight-to-VRAM math, the fit rules, and how live prices are normalized: full methodology.
More Llama 3.3 models
- Llama-3_3-Nemotron-Super-49B-v1_5-FP8 (49.9B, F8_E4M3)
- Llama-3_3-Nemotron-Super-49B-v1 (49.9B, BF16)
- Llama-3_3-Nemotron-Super-49B-v1_5 (49.9B, BF16)
Lineage
Based on Llama-3.1-70B.
Quantized and fine-tuned versions
Also published as: unsloth/Llama-3.3-70B-Instruct. Identical copies of this model under another account.
Alternatives at this size
Other models for text-generation within about a third of Llama-3.3-70B-Instruct's 70.6B parameters, from other model lines.
- Qwen-72B (72.3B, BF16)
- Kimi-Linear-48B-A3B-Instruct (49.1B, BF16)
- Hunyuan-A13B-Instruct (80.4B, BF16)
- Apertus-70B-Instruct-2509 (70.6B, BF16)
- Le_Triomphant-ECE-TW3 (72.3B, BF16)
More on Llama-3.3-70B-Instruct
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.
Best chat and assistants models: how Llama-3.3-70B-Instruct ranks against the rest.
Compare Llama-3.3-70B-Instruct with Qwen-72B, Qwen3-Coder-Next, Qwen2.5-72B-Instruct, Phi-3.5-MoE-instruct, Qwen2.5-72B.
Related reading: RTX A5000 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.
