all-MiniLM-L6-v2 on NVIDIA Grace Blackwell Ultra GB300
Yes — Grace Blackwell Ultra GB300 runs all-MiniLM-L6-v2 excellently at FP16 — 3000 tok/s. 288 GB VRAM with plenty of headroom.
Model Size
23M
Device VRAM
288 GB
Bandwidth
8000 GB/s
Quantization
FP16
Performance by Quantization
OwnRig currently has one published compatibility entry for all-MiniLM-L6-v2 on NVIDIA Grace Blackwell Ultra GB300 at FP16. This is the best supported pairing we can stand behind today.
| Quantization | Speed | TTFT | Fits in VRAM | Rating | Confidence |
|---|---|---|---|---|---|
| FP16 | 3000 tok/s | 2ms | ✓ Yes | Excellent | estimated |
Notes
FP16
Tiny embedding model. 288GB headroom. Practically instant.
About all-MiniLM-L6-v2
all-MiniLM-L6-v2 (23M) is a embeddings, ai building model. Ultra-lightweight embedding model at 23M params. Fastest option for local embeddings. Lower quality than nomic-embed but practically free in VRAM. Ideal when running alongside large coding models and every MB of VRAM matters. The tiny model for builders who need RAG without the VRAM cost.
View all all-MiniLM-L6-v2 hardware options →About NVIDIA Grace Blackwell Ultra GB300
NVIDIA Grace Blackwell Ultra GB300 has 288 GB at 8000 GB/s. Street price: $30,000.
See all models NVIDIA Grace Blackwell Ultra GB300 can run →Estimate method: Estimated from GB300 bandwidth ratio vs RTX 4090. Reference hardware source: dell.com (2026-03-23)
Performance varies by driver version, inference engine, quantization method, context length, and system configuration. Figures shown are estimates based on community benchmarks and may not reflect your exact setup. Product names are trademarks of their respective owners. OwnRig is independent and not affiliated with any hardware or AI model provider.