GigaChat Lightning 10B on NVIDIA Grace Blackwell Ultra GB300
Yes — Grace Blackwell Ultra GB300 runs GigaChat Lightning 10B excellently at Q8_0 — 320 tok/s. 288 GB VRAM with plenty of headroom.
Model Size
10B
Device VRAM
288 GB
Bandwidth
8000 GB/s
Quantization
Q8_0
Performance by Quantization
OwnRig currently has one published compatibility entry for GigaChat Lightning 10B on NVIDIA Grace Blackwell Ultra GB300 at Q8_0. This is the best supported pairing we can stand behind today.
| Quantization | Speed | TTFT | Fits in VRAM | Rating | Confidence |
|---|---|---|---|---|---|
| Q8_0 | 320 tok/s | 19ms | ✓ Yes | Excellent | estimated |
Notes
Q8_0
Q8_0 fits in 288GB with 277GB headroom. Small MoE model; 1.8B active per token keeps decode speed high once loaded.
About GigaChat Lightning 10B
GigaChat Lightning 10B (10B) is a chat, multi-purpose model. Compact MoE model with 10B total parameters and around 1.8B active per token. Interesting efficiency play for chat workloads that want MoE-style speed without the memory cost of larger expert models.
View all GigaChat Lightning 10B 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 small-model MoE benchmarks and bandwidth scaling. Reference hardware source: github.com (2026-03-24)
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.