NVIDIA · Embeddings and rerank · 7.5B parameters · other
llama-embed-nemotron-8b
Needs 9 GB at FP8 / INT8. The smallest real configuration today is 1× RTX 3060 at Vast.ai, $0.04 per hour, about $26 a month at 730 hours.
Parameters7.5Bdense
Context128k tokensmaximum published
Checkpoint15 GBBF16, as published
Licenseothercheck before commercial use
Released—documented date
Smallest configuration · FP8 / INT8
1× RTX 306012 GB VRAM · 75% used
- Per hour
- $0.04
- Per month
- $26
- Cheapest at
Vast.ai
Memory · 3 precisions · 4 context lengthsHow much VRAM it needs
Weights plus runtime overhead, then the KV cache that grows with the context you serve.
| Precision | Weights | Weights + headroom | 4k context | 32k context | 128k context |
|---|---|---|---|---|---|
| BF16published checkpoint | 15 GB | 18 GB | 19 GB | 23 GB | 36 GB |
| FP8 / INT8estimated | 8 GB | 9 GB | 10 GB | 14 GB | 27 GB |
| INT4 / MXFP4estimated | 5 GB | 6 GB | 6 GB | 10 GB | 23 GB |
| KV cacheone sequence, FP16 cache, included above | — | — | 1 GB | 5 GB | 18 GB |
Cross-checked on Hugging Face Oct 28, 2026: 7.50B parameters in the safetensors headers, 15.0 GB of weight files, attention layout from config.json (32 layers, 8 KV heads × 128). Method: weights as published at the native precision, verified on Hugging Face 2026-10-28; other precisions estimated at 1 (8-bit) and 0.6 (4-bit) bytes per parameter; × 1.2 runtime headroom; KV cache = 4096 bytes per token per layer × 32 full-attention layers (FP16 cache, one sequence).config.json ↗
Smallest configuration per GPU · 110 GPUs · cheapest eligible on-demand offer per GPU count, every provider on fileWhat it costs to run, GPU by GPU
The fewest GPUs of one type whose total memory fits the model at short context, priced from the cheapest eligible on-demand offer for exactly that count (45-day window, newest reading per provider), with the next providers beside it. Monthly = 730 hours.
| GPU | Smallest config | VRAM used | $ / hour | $ / month | Cheapest at |
|---|---|---|---|---|---|
| RTX 3060Best fit | 1× 12 GB12 GB total | 75% | $0.04 | $26 | Vast.aiTrust 78 · read Nov 2 · receipt · also Clore.ai $0.05, gcube $0.05 |
| RTX A2000 | 1× 12 GB12 GB total | 75% | $0.04 | $26 | Vast.aiTrust 78 · read Nov 2 · receipt · also gcube $0.09, Runpod $0.12 |
| T4 | 1× 16 GB16 GB total | 56% | $0.04 | $29 | Vast.aiTrust 78 · read Nov 2 · receipt · also iwinv $0.20, Theta EdgeCloud $0.20 |
| Tesla P40 | 1× 24 GB24 GB total | 38% | $0.04 | $30 |  Clore.aiTrust 49 · read Nov 2 · receipt · also gcube $0.07, Vast.ai $0.11 |
| GTX Titan X | 1× 12 GB12 GB total | 75% | $0.04 | $31 | Vast.aiTrust 78 · read Nov 2 · receipt |
| Titan Xp | 1× 12 GB12 GB total | 75% | $0.04 | $31 | Vast.aiTrust 78 · read Nov 2 · receipt |
| Tesla P100 | 1× 16 GB16 GB total | 56% | $0.04 | $33 | Vast.aiTrust 78 · read Nov 2 · receipt · also gcube $0.07 |
| Quadro P2000 | 2× 5 GB10 GB total | 90% | $0.06 | $41 | Vast.aiTrust 78 · read Nov 2 · receipt |
| Tesla V100 16GB | 1× 16 GB16 GB total | 56% | $0.06 | $46 | Clore.aiTrust 49 · read Nov 2 · receipt · also Vast.ai $0.08, CloudRift $0.28 |
| GTX 1080 Ti | 1× 11 GB11 GB total | 82% | $0.06 | $47 | Vast.aiTrust 78 · read Nov 2 · receipt · also Edgevana $0.08 |
| RTX 2080 Ti | 1× 11 GB11 GB total | 82% | $0.07 | $49 | |
| Tesla V100 32GB | 1× 32 GB32 GB total | 28% | $0.07 | $50 | Vast.aiTrust 78 · read Nov 2 · receipt · also Cyfuture AI $0.57, E2E Networks $1.50 |
| RTX PRO 2000 | 1× 16 GB16 GB total | 56% | $0.07 | $51 | Clore.aiTrust 49 · read Nov 2 · receipt · also gcube $0.16 |
| RTX A4000 | 1× 16 GB16 GB total | 56% | $0.07 | $53 | Vast.aiTrust 78 · read Nov 2 · receipt · also Edgevana $0.10, gcube $0.10 |
| Tesla M40 | 1× 24 GB24 GB total | 38% | $0.07 | $54 | Clore.aiTrust 49 · read Nov 2 · receipt · also gcube $0.07 |
| RTX 3080 | 1× 10 GB10 GB total | 90% | $0.07 | $55 | |
| RTX 3060 Ti | 2× 8 GB16 GB total | 56% | $0.08 | $61 | Vast.aiTrust 78 · read Nov 2 · receipt · also Clore.ai $0.12, AtmosCompute $0.34 |
| RTX 5060 Ti | 1× 16 GB16 GB total | 56% | $0.08 | $61 | Clore.aiTrust 49 · read Nov 2 · receipt · also Vast.ai $0.09, Theta EdgeCloud $0.12 |
| RTX 4070 | 1× 12 GB12 GB total | 75% | $0.08 | $61 | Clore.aiTrust 49 · read Nov 2 · receipt · also Vast.ai $0.11, gcube $0.14 |
| GTX 1660 | 2× 6 GB12 GB total | 75% | $0.09 | $64 | Vast.aiTrust 78 · read Nov 2 · receipt |
| RTX 3090 | 1× 24 GB24 GB total | 38% | $0.09 | $66 | |
| RTX 4070 Super | 1× 12 GB12 GB total | 75% | $0.10 | $70 | Vast.aiTrust 78 · read Nov 2 · receipt · also Clore.ai $0.13, gcube $0.15 |
| GTX 1070 | 2× 8 GB16 GB total | 56% | $0.10 | $70 | Vast.aiTrust 78 · read Nov 2 · receipt |
| GTX 1080 | 2× 8 GB16 GB total | 56% | $0.10 | $76 | Vast.aiTrust 78 · read Nov 2 · receipt |
| The 24 cheapest of 100 GPU types that fit. Every GPU, any precision and context, in the finder | |||||
Throughput estimates · Oct 4, 2026Tokens per second, and what a million costs
Embeddings and rerank. Speed estimated from memory bandwidth and model size, never measured. The cost per 1B tokens is an estimate at the prices read on Oct 4, 2026; it is not recomputed here.
| GPU | Cards | Precision | tokens / s | per 1B tokens · estimate |
|---|---|---|---|---|
| RTX 4090 | 1 | FP16 | 7695 | $8.30 |
| RTX 5090 | 1 | FP16 | 9770 | $13.60 |
| L40S | 1 | FP16 | 8441 | $18.10 |
| MI300X | 1 | FP16 | 24381 | $18.80 |
| H100 PCIe | 1 | FP16 | 17628 | $21.30 |
| B200 | 1 | FP16 | 52465 | $22.00 |
| H100 SXM | 1 | FP16 | 23061 | $22.30 |
| RTX PRO 6000 | 1 | FP16 | 11729 | $27.20 |
| H200 | 1 | FP16 | 23061 | $30.00 |
| A100 80GB | 1 | FP16 | 7275 | $36.30 |
cards, precision and speed estimated from memory bandwidth and model size; cost recomputed by the site from verified prices · cheapest per output at the prices of Oct 4, 2026: RTX 4090Why estimates, and their limits →
llama-embed-nemotron-8b · related