141 GB VRAM (+45)4800 GB/s (+800)
A
Raises estimated decode speed by about 56%.97.4 tok/s decode
Raises estimated decode speed by about 56%.
Adds memory headroom for longer context windows and future model growth.
~$30,000 MSRP
Can Mixtral 8x22B run on NVIDIA H20 96GB?
YES — With Offload
B66Good
○Estimated from fit model
Mixtral 8x22B needs ~99.9 GB VRAM. NVIDIA H20 96GB has 96.0 GB. With Q4_K_M quantization, expect ~58 tok/s.
Runtime: llama.cppCapacity: OffloadBandwidth: HighStack: StandardBottleneck: Host offload
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
Q4_K_M (Medium quality) — 99.9 GB, 62.5 tok/s, Runs with offload (needs ~3.4 GB host RAM)
99.9 GB required96.0 GB available
104% VRAM needed
3.9 GB over capacity — needs offload or smaller quantization
Fit status
Runs with offload (needs ~3.4 GB host RAM)
Decode
62.5 tok/s
TTFT
3098 ms
Safe context
4K
Memory
99.9 GB / 96.0 GB
Memory breakdown
Weights86.0 GB
KV Cache3.4 GB
Runtime0.9 GB
Headroom9.6 GB
See how fast it feels
See how fast it feelsMixtral 8x22B on NVIDIA H20 96GB
1st promptCold start — includes initialization
>Solve: If a train travels at 60 mph for 2.5 hours, how far does it go?
Let me work through this step by step.
Given information:
- Speed = 60 mph (miles per hour)
- Time = 2.5 hours
Formula: Distance = Speed x Time
Calculation:
Distance = 60 mph x 2.5 h
= 150 miles
The train travels 150 miles.
Double-check: at 60 mph the train covers 60 miles each hour. In half an hour it covers 30 miles. So in 2.5 hours: 60 + 60 + 30 = 150 miles. Confirmed.
2nd promptWarm — model ready, no init overhead
>What if the train stops for 20 minutes halfway?
The stop doesn't change the total distance covered while moving, but it affects total trip time.
Breakdown:
- First half: 1.25 hours at 60 mph = 75 miles
- Stop: 20 minutes (no distance)
- Second half: 1.25 hours at 60 mph = 75 miles
Total distance: 150 miles (same)
Total time: 2.5 h + 0.33 h = 2.83 hours
Average speed: 150 / 2.83 = 53 mph
The distance stays the same but average speed drops to 53 mph because of the stop.
Estimated: 62.5 tok/s decode · 3.1s TTFT (warm) · 156 tok/s prefill
What limits this setup
This setup is broadly balanced for this model.
Very little memory headroom
You can run the model, but there is not much room left for longer context, bigger batches, extra apps, or future model updates.
Best improvement path
Buy headroom, not only minimum fit
A slightly larger memory tier gives you safer context growth and makes the recommendation more future-proof.
Performance by workload
| Workload | Grade | Fit | Decode | TTFT | Context |
|---|---|---|---|---|---|
| Chat | B | Runs with offload | 59.4 tok/s | 1779 ms | 4K |
| Coding | B | Runs with offload | 57.7 tok/s | 3356 ms | 4K |
| Agentic Coding | B | Runs with offload | 54.6 tok/s | 5160 ms | 4K |
| Reasoning | B | Runs with offload | 57.7 tok/s | 3966 ms | 4K |
| RAG | B | Runs with offload | 54.6 tok/s | 6450 ms | 4K |
Quantization options
How Mixtral 8x22B (141B params) fits at each quantization level on NVIDIA H20 96GB (96.0 GB usable).
| Quant | Bits | VRAM | Quality | Fit |
|---|---|---|---|---|
Q2_K | 2 | 55.0 GB | Low | B61 |
Q3_K_SBest for your GPU | 3 | 69.1 GB | Low | B61 |
Get started
Copy-paste commands to run Mixtral 8x22B on your machine.
Run
ollama run mixtral:8x22bUpgrade options
Hardware that runs Mixtral 8x22B well
141 GB VRAM (+45)4800 GB/s (+800)
A
Raises estimated decode speed by about 56%.97.4 tok/s decode
Raises estimated decode speed by about 56%.
Adds memory headroom for longer context windows and future model growth.
~$30,000 MSRP
180 GB VRAM (+84)8000 GB/s (+4000)
B
Raises estimated decode speed by about 160%.162.4 tok/s decode
Raises estimated decode speed by about 160%.
Adds memory headroom for longer context windows and future model growth.
~$30,000 MSRP