Can RTX 3090 24GB run Gemma 4 12B?
Measured results for Gemma 4 12B (11.96B dense) on RTX 3090 24GB, plus what fits and the commands to run it.
Short answer
Yes. Gemma 4 12B Q8_0 fits fully on RTX 3090 24GB at 16k context, using 13.7 of 23.0 GB usable, and generates about 57.2 tokens/s in our own run. For Q8_0, our bandwidth-only estimate was ~39 tok/s against a measured median of ~57 (32% too pessimistic).
Measured generation speed by quantization
| Quant | Generation tok/s | Range | Prompt tok/s | Runs | Origin |
|---|---|---|---|---|---|
| Q4 class (median) | 86.5 | 86.5–86.5 | 3272 | 1 | includes our runs |
| Q8 class (median) | 57.2 | 57.2–57.2 | 3408 | 1 | includes our runs |
| Q4_K_M (our run, depth 0) | 86.5 | 3 repeats | 3272 | 3 | own measurement |
| Q8_0 (our run, depth 0) | 57.2 | 3 repeats | 3408 | 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 GeForce RTX 3090 (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 | 86.5 | 100% | 3272 |
| Q4_K_M | 4k | 82.7 | 96% | 2961 |
| Q4_K_M | 16k | 80.2 | 93% | 2439 |
| Q8_0 | 0 | 57.2 | 100% | 3408 |
| Q8_0 | 4k | 55.5 | 97% | 3080 |
| Q8_0 | 16k | 54.3 | 95% | 2523 |
Memory and context limits
Q8_0 weights are 13.1 GB; the 16k f16 KV cache adds 0.60 GB. Largest context that fits in 23.0 GB usable:
| KV cache type | Max context |
|---|---|
| f16 | 256k |
| q8_0 | 256k |
| q4_0 | 256k |
Native limit 256k. Full KV cache table for Gemma 4 12B.
Copy-paste commands
The advisor's pick for this hardware: Q8_0, 16k context.
llama-server -hf unsloth/gemma-4-12b-it-GGUF:Q8_0 -c 16384 -fa on --jinja -ngl 99
OLLAMA_CONTEXT_LENGTH=16384 ollama run hf.co/unsloth/gemma-4-12b-it-GGUF:Q8_0
FAQ
How fast is Gemma 4 12B on RTX 3090 24GB?
About 57.2 tokens/s for generation (Q8_0 on llama.cpp CUDA, 128-token generation), 3408 tokens/s for prompt processing. Speed drops as the context fills; see the depth table.
How much memory does Gemma 4 12B need on RTX 3090 24GB?
The advisor's pick, Q8_0, is 13.1 GB of weights plus 0.60 GB of KV cache at 16k context (13.7 GB), against 23.0 GB usable.
What is the maximum context for Gemma 4 12B on RTX 3090 24GB?
With an f16 KV cache about 256k tokens, 256k with q8_0 and 256k with q4_0, bounded by the model's own limit.
Related
All models on RTX 3090 24GB · Gemma 4 12B on all hardware · VRAM tier · all measured pages · leaderboard
Compare: RTX 3090 24GB vs RTX 4090 24GB · RTX 3090 24GB vs RTX 5090 32GB · RTX 3090 24GB vs AMD RX 7900 XTX 24GB