Can RTX PRO 6000 96GB run gpt-oss 120B?
Measured results for gpt-oss 120B (116.83B MoE, 5.1B active) on RTX PRO 6000 96GB, plus what fits and the commands to run it.
Short answer
Yes. gpt-oss 120B Q8_0 fits fully on RTX PRO 6000 96GB at 16k context, using 64.0 of 95.0 GB usable, and generates about 260.8 tokens/s in our own run. As the context fills it slows down: 231.4 tok/s with 16k tokens already in context (89% of the empty-context speed), measured. For Q8_0, our bandwidth-only estimate was ~356 tok/s against a measured median of ~261 (37% too optimistic).
Measured generation speed by quantization
| Quant | Generation tok/s | Range | Prompt tok/s | Runs | Origin |
|---|---|---|---|---|---|
| Q8 class (median) | 260.8 | 260.8–260.8 | 9342 | 1 | includes our runs |
| Q4 class (median) | 236.6 | 196.3–276.9 | 7268 | 2 | includes our runs |
| Q4_K_M (our run, depth 0) | 276.9 | 3 repeats | 9017 | 3 | own measurement |
| Q8_0 (our run, depth 0) | 260.8 | 3 repeats | 9342 | 3 | own measurement |
Generation = tg128 (128 generated tokens, short context). Public rows are medians of community llama.cpp results (sources); our own runs use llama-bench b11658 on a rented GPU.
Speed vs context depth measured
Our own llama-bench runs on RTX PRO 6000 Blackwell Workstation (flash attention on, 3 repeats): generation and prompt speed with the given number of tokens already in the context.
| Quant | Context depth | Generation tok/s | vs depth 0 | Prompt tok/s |
|---|---|---|---|---|
| Q4_K_M | 0 | 276.9 | 100% | 9017 |
| Q4_K_M | 4k | 249.9 | 90% | 8761 |
| Q4_K_M | 16k | 242.5 | 88% | 7708 |
| Q8_0 | 0 | 260.8 | 100% | 9342 |
| Q8_0 | 4k | 237.1 | 91% | 8968 |
| Q8_0 | 16k | 231.4 | 89% | 7870 |
Alternatives
Less memory that still fits gpt-oss 120B
Smaller memory usually means a cheaper card or Mac (we do not list prices).
- Mac Studio (2025) M3 Ultra 96GB: Q8_0, 67 GB usable, ~92 tok/s estimated
- Mac Studio M5 Max 128GB: Q8_0, 89 GB usable, ~107 tok/s estimated
- MacBook Pro 14/16 M5 Max 128GB: Q8_0, 89 GB usable, ~107 tok/s estimated
Other models on RTX PRO 6000 96GB, similar size
- GLM-4.5-Air: Q5_K_M, fits fully, ~121 tok/s (scaled from measured)
- Nemotron 3.5 Super VL 120B-A12B (hardware page): Q4_K_M, fits fully, ~162 tok/s (scaled from measured)
- Llama 3.3 70B (hardware page): Q8_0, fits fully, ~15 tok/s (estimated)
- Qwen3.6 35B-A3B (hardware page): Q8_0, fits fully, ~247 tok/s (measured)
Memory and context limits
Q8_0 weights are 63.4 GB; the 16k f16 KV cache adds 0.61 GB. Largest context that fits in 95.0 GB usable:
| KV cache type | Max context |
|---|---|
| f16 | 128k |
| q8_0 | 128k |
| q4_0 | 128k |
Native limit 128k. Full KV cache table for gpt-oss 120B.
Copy-paste commands
The advisor's pick for this hardware: Q8_0, 16k context.
llama-server -hf unsloth/gpt-oss-120b-GGUF:Q8_0 -c 16384 -fa on --jinja -ngl 99
OLLAMA_CONTEXT_LENGTH=16384 ollama serve ollama run hf.co/unsloth/gpt-oss-120b-GGUF:Q8_0
FAQ
How fast is gpt-oss 120B on RTX PRO 6000 96GB?
About 260.8 tokens/s for generation (Q8_0 on llama.cpp CUDA, 128-token generation), 9342 tokens/s for prompt processing. Speed drops as the context fills; see the depth table.
How much memory does gpt-oss 120B need on RTX PRO 6000 96GB?
The advisor's pick, Q8_0, is 63.4 GB of weights plus 0.61 GB of KV cache at 16k context (64.0 GB), against 95.0 GB usable.
Is the gpt-oss 120B speed on RTX PRO 6000 96GB measured or estimated?
Measured: our own llama-bench run on a rented RTX PRO 6000 Blackwell Workstation, including runs with 4k to 16k tokens already in context.
What can I use instead of RTX PRO 6000 96GB for gpt-oss 120B?
With less memory, Mac Studio (2025) M3 Ultra 96GB (~92 tok/s), Mac Studio M5 Max 128GB (~107 tok/s), MacBook Pro 14/16 M5 Max 128GB (~107 tok/s) also fit it fully.
What is the maximum context for gpt-oss 120B on RTX PRO 6000 96GB?
With an f16 KV cache about 128k tokens, 128k with q8_0 and 128k with q4_0, bounded by the model's own limit.
Related
All models on RTX PRO 6000 96GB · gpt-oss 120B on all hardware · VRAM tier · all measured pages · leaderboard