Can RTX 3090 run gpt-oss 120B?

Last updated: 31 Oct 2026

Probably only partly: the weights are larger than its memory, so part of the model likely ran from system RAM. Measured 18.1 tok/s generation with part of the model offloaded to system RAM (llama.cpp CUDA b11804, rest mmapped from NVMe page cache, Q4_K_M (58.5 GiB, larger than VRAM + RAM); source).

Every measured result for gpt-oss 120B on RTX 3090

Build	Quant / setup	Engine	tok/s
RTX 3090 + Ryzen 5 5600X + 32 GB DDR4: gpt-oss 20B at 211 tok/s, Qwen3.6 35B-A3B at 172, even gpt-oss 120B at 18 via NVMe paging (llm-benchmark.de)	Q4_K_M (58.5 GiB, larger than VRAM + RAM)	llama.cpp CUDA b11804, rest mmapped from NVMe page cache	18.1 CPU offload	source

Memory: RTX 3090 has 24 GB. gpt-oss 120B weights are about 65 GB in its native format (OpenAI model card: 117B total, 5.1B active). Arithmetic, not a measurement; long contexts need more. Check other quants in the VRAM calculator.

Speeds are decode (generation) tokens/sec as reported by each source; quant, engine and context differ, so small gaps are noise.

All measured speeds for gpt-oss 120B: best home rig for gpt-oss 120B.

gpt-oss 120B on other hardware: Apple M2 Ultra · DGX Spark (GB10) · RTX Pro 6000 (Blackwell, 96 GB) · Strix Halo (Radeon 8060S)

Other models measured on RTX 3090: DeepSeek R1 671B · Flint 40B-A2.5B · GLM 4.7 Flash · Gemma 4 31B · Nemotron 3.5 Bolt 74B-A9B · Nemotron 3.5 Lightning 30B-A3B · Qwen3.6 27B · Qwen3.6 35B-A3B · Qwen3.8 27B · gpt-oss 20B

Not enough VRAM? Rent before you buy. Run gpt-oss 120B on a bigger GPU on a rented 24–96 GB cloud GPU by the hour and see the real speed before spending €1,000+. Try it on Runpod
Referral link: if you sign up with Google and add $10, Runpod gives you and RigBase bonus credit (terms). It never affects our data.