Mac Studio M3 Ultra 256GBBudget pick
256 GB Unified (+160)
A
Adds memory headroom for longer context windows and future model growth.4.7 tok/s decode
Adds memory headroom for longer context windows and future model growth.
~$6,999 MSRP
Can mxbai Embed Large run on NVIDIA GH200 96GB?
YES — Runs Great
B70Good
○Estimated from fit model
mxbai Embed Large needs ~12.5 GB VRAM. NVIDIA GH200 96GB has 96.0 GB. With F16 quantization, expect ~5 tok/s.
Runtime: OllamaCapacity: RoomyBandwidth: HighStack: BasicBottleneck: Memory bandwidth
Share:
Operating mode
Choose the run profile you care about
Interactive favors responsiveness, while light API and scale-out lean harder on serving readiness. The fit stays the same, but the recommendation lens changes.
Current mode
Balanced
Balanced for general local use. Keeps the ranking neutral across personal and serving workflows.
Select quantization to explore
F16 (Maximum quality) — 13.0 GB, 4.7 tok/s, Runs well
13.0 GB required96.0 GB available
14% VRAM used
Fit status
Runs well
Decode
4.7 tok/s
TTFT
41279 ms
Safe context
512
Memory
13.0 GB / 96.0 GB
Memory breakdown
Weights0.7 GB
KV Cache1.5 GB
Runtime1.2 GB
Headroom9.6 GB
See how fast it feels
See how fast it feelsmxbai Embed Large on NVIDIA GH200 96GB
1st promptCold start — includes initialization
>What is local AI inference?
Local AI inference means running an AI language model directly on your own hardware — your laptop, desktop, or server — instead of sending requests to a remote cloud API.
When you run inference locally the model weights are loaded into your GPU or unified memory. Each token you generate requires reading those weights from memory, so memory bandwidth is the main bottleneck for decode speed.
Key benefits of running locally:
- Full privacy: your prompts never leave your machine
- No per-token cost or rate limits
- Works offline once the model is downloaded
- Latency depends only on your hardware
2nd promptWarm — model ready, no init overhead
>How much VRAM do I need?
It depends on the model size and quantization level. A rough rule of thumb:
Model size Q4 (4-bit) Q8 (8-bit) FP16
7B params ~4.3 GB ~7.5 GB ~14 GB
13B params ~7.9 GB ~13.9 GB ~26 GB
70B params ~42.7 GB ~74.9 GB ~140 GB
Most people use 4-bit quantization (Q4_K_M) which gives 90-95% of full quality at a fraction of the memory. A 24 GB GPU can comfortably run most 7B-13B models.
Estimated: 4.7 tok/s decode · 41.3s TTFT (warm) · 12 tok/s prefill
What limits this setup
This model fits, but memory bandwidth is the part holding decode speed back.
Throughput will feel slow
Estimated decode speed is only 4.7 tok/s, so this is more of a technical fit than a comfortable daily-driver setup.
Best improvement path
Prioritize bandwidth, not only capacity
If this workload feels slow, the next useful step is often a GPU tier with materially faster memory bandwidth rather than only a small bump in capacity.
Performance by workload
| Workload | Grade | Fit | Decode | TTFT | Context |
|---|---|---|---|---|---|
| Chat | B | Runs well | 4.7 tok/s | 22516 ms | 512 |
| Coding | B | Runs well | 4.7 tok/s | 41279 ms | 512 |
| Agentic Coding | B | Runs well | 4.7 tok/s | 60043 ms | 512 |
| Reasoning | B | Runs well | 4.7 tok/s | 48785 ms | 512 |
| RAG | B | Runs well | 4.7 tok/s | 75053 ms | 512 |
Quantization options
How mxbai Embed Large (0.33500000834465027B params) fits at each quantization level on NVIDIA GH200 96GB (96.0 GB usable).
| Quant | Bits | VRAM | Quality | Fit |
|---|---|---|---|---|
Q2_K | 2 | 0.1 GB | Low | A73 |
Q3_K_S | 3 | 0.2 GB | Low | A73 |
NVFP4 | 4 |
Get started
Copy-paste commands to run mxbai Embed Large on your machine.
Run
ollama run mxbai-embed-largeUpgrade options
