Skip to content

VRAM reference

Llama 3.1 70B on an RTX 5090

No fine-tuning method fits Llama 3.1 70B on one RTX 5090 (32 GB) at 2,048 tokens. 4-bit inference: no, the weights alone are 40 GB.

By Will Schulz · Checked 15 November 2026 · Prices as of November 2026

Verdict by method

Fine-tuningMemory at 2K tokensOne RTX 5090?Longest sequence (batch 1)Largest batch at 2KLargest LoRA rank
QLoRA (4-bit base)48 GBNo———
LoRA (16-bit base)150 GBNo———
Full fine-tuning1,134 GBNo——n/a
InferenceMemory at 8KOne RTX 5090?Longest contextConcurrent 4K sequences
16-bit (BF16/FP16)144 GBNo——
8-bit73 GBNo——
4-bit42 GBNo——
  • QLoRA: no, it needs about 48 GB at 2K tokens, more than the 32 GB available.
  • LoRA: no, it needs about 150 GB at 2K tokens, more than the 32 GB available.
  • Full fine-tuning: no, it needs about 1,134 GB at 2K tokens, more than the 32 GB available.

Where the memory goes

At 2,048 tokens per sequence, batch 1, LoRA rank 16, gradient checkpointing on
MethodWeightsTraining stateActivationsTotalOn one RTX 509080 GB GPUs needed
QLoRA (4-bit base)40 GB3.3 GB5.6 GB48 GBDoesn't fit1
LoRA (16-bit base)141 GB3.3 GB5.6 GB150 GBDoesn't fit3
Full fine-tuning141 GB988 GB5.6 GB1,134 GBDoesn't fit16
One sequence, fp16 KV cache
PrecisionWeightsKV cache at 8KTotalAt 128K contextOn one RTX 5090
16-bit (BF16/FP16)141 GB2.7 GB144 GB184 GBDoesn't fit
8-bit71 GB2.7 GB73 GB114 GBDoesn't fit
4-bit40 GB2.7 GB42 GB83 GBDoesn't fit

The cheapest way to run it

Renting an RTX 5090 costs $1.39 an hour on Runpod Secure Cloud ($0.69 on Runpod Community Cloud). Since no fine-tuning method fits, the table shows the cheapest hardware that does.

Job (default settings)MemoryCheapest rental that fitsPer hour
QLoRA fine-tuning48 GBA100 80GB$2.19
LoRA fine-tuning150 GB4 × L40S$4.36
4-bit inference42 GBL40S$1.09
16-bit inference144 GB2 × A100 80GB$4.38

On-demand rates seen on 15 November 2026: Runpod Secure Cloud, Lambda for the A100 40GB. Consumer cards count as single GPUs only; data-centre GPUs up to eight per node.

Try other settings

JobMemoryRTX 5090 (32 GB)Smallest single GPU
QLoRA (4-bit base)2K tokens × batch 1, rank 1648 GB2 GPUsA100 80GB (80 GB)
LoRA (16-bit base)2K tokens × batch 1, rank 16150 GB6 GPUsB200 (180 GB)
Full fine-tuning2K tokens × batch 11,134 GB40 GPUsNone: needs several
Inference, 16-bit (BF16/FP16)8K-token context × 1 sequence144 GB5 GPUsB200 (180 GB)
Inference, 8-bit8K-token context × 1 sequence73 GB3 GPUsH200 (141 GB)
Inference, 4-bit8K-token context × 1 sequence42 GB2 GPUsL40S (48 GB)

A job fits when the estimate is within 90% of the card's memory. Training assumes gradient checkpointing and FlashAttention; multi-GPU counts assume the memory splits evenly (FSDP, ZeRO-3 or tensor parallelism).

How we estimate it

Every figure comes from the model's published architecture (layers, hidden size, attention and KV heads, vocabulary) rather than a rule of thumb, using the same functions as the calculator above. Sizes are in GB (10⁹ bytes).

  • Full fine-tuning: 16 bytes per parameter for mixed-precision Adam (bf16 weights and gradients, fp32 master weights and two moments), as in the ZeRO paper.
  • LoRA: frozen bf16 weights (2 bytes per parameter) plus 16 bytes per adapter parameter, rank 16 on every linear layer. For mixture-of-experts models, adapters go on attention only and the experts stay frozen.
  • QLoRA: NF4 weights with double quantisation (4.127 bits per parameter); embeddings and the output head stay in bf16, as bitsandbytes leaves them.
  • Activations: gradient checkpointing and FlashAttention. Each layer's bf16 input is kept, one layer is recomputed at a time, and the output logits take 8 bytes per vocabulary entry per token (fp32 values and gradients). Chunked cross-entropy, as in Unsloth or Liger kernels, cuts that last term.
  • Inference: 2, 1 or about 0.56 bytes per weight (4-bit with group-wise scales), plus an fp16 KV cache of 2 × layers × KV heads × head dim × 2 bytes per token; sliding-window layers only cache their window.
  • Fits means the estimate is within 90% of the card's listed memory; the rest covers the CUDA context, kernels and fragmentation. Multi-GPU counts assume memory splits evenly.

Speed estimates use RTX 5090's dense BF16 peak of 209.5 TFLOPS, about 4 FLOPs per active parameter per token for LoRA and 40% utilisation; see fine-tuning and training costs for the method.

Frequently asked questions

Can I fine-tune Llama 3.1 70B on an RTX 5090?
QLoRA: no, it needs about 48 GB at 2K tokens, more than the 32 GB available. LoRA: no, it needs about 150 GB at 2K tokens, more than the 32 GB available. Full fine-tuning: no, it needs about 1,134 GB at 2K tokens, more than the 32 GB available.
Can I run Llama 3.1 70B on an RTX 5090?
16-bit (BF16/FP16) inference: no, the weights alone are 141 GB. 8-bit inference: no, the weights alone are 71 GB. 4-bit inference: no, the weights alone are 40 GB.

Sources

  1. [1]Hugging Face. meta-llama/Llama-3.1-70B-Instruct config.json
  2. [2]Meta. meta-llama/Llama-3.1-70B-Instruct model card
  3. [3]Meta. The Llama 3 Herd of Models (arXiv:2407.21783)
  4. [4]NVIDIA. RTX 5090 specifications
  5. [5]Runpod. GPU pricing (updated 12 November 2026)
  6. [6]Lambda. GPU cloud pricing
  7. [7]Dettmers et al., 2023. QLoRA: Efficient Finetuning of Quantized LLMs (arXiv:2305.14314)
  8. [8]Hu et al., 2021. LoRA: Low-Rank Adaptation of Large Language Models (arXiv:2106.09685)
  9. [9]Rajbhandari et al., 2019. ZeRO: Memory Optimizations Toward Training Trillion Parameter Models (arXiv:1910.02054)
  10. [10]Korthikanti et al., 2022. Reducing Activation Recomputation in Large Transformer Models (arXiv:2205.05198)

Measure before you rent the GPU.

Prompt tuning and model search in Kojen run on hosted models against your own test cases, so you know whether fine-tuning Llama 3.1 70B is worth the GPU time.

Start training free