Can RTX 5090 32GB run Gemma 4 26B-A4B?
Measured results for Gemma 4 26B-A4B (25.81B MoE, 3.8B active) on RTX 5090 32GB, plus what fits and the commands to run it.
Short answer
Yes. Gemma 4 26B-A4B Q8_0 fits fully on RTX 5090 32GB at 16k context, using 27.4 of 31.0 GB usable, and generates about 217.5 tokens/s in our own run. As the context fills it slows down: 201.7 tok/s with 16k tokens already in context (93% of the empty-context speed), measured. For Q8_0, our bandwidth-only estimate was ~249 tok/s against a measured median of ~217 (15% too optimistic).
Measured generation speed by quantization
| Quant | Generation tok/s | Range | Prompt tok/s | Runs | Origin |
|---|---|---|---|---|---|
| Q4 class (median) | 231.4 | 231.4–231.4 | 11432 | 1 | includes our runs |
| Q8 class (median) | 217.5 | 217.5–217.5 | 11322 | 1 | includes our runs |
| Q4_K_M (our run, depth 0) | 231.4 | 3 repeats | 11432 | 3 | own measurement |
| Q8_0 (our run, depth 0) | 217.5 | 3 repeats | 11322 | 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 5090 (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 | 231.4 | 100% | 11432 |
| Q4_K_M | 4k | 216.4 | 94% | 10289 |
| Q4_K_M | 16k | 213 | 92% | 9262 |
| Q8_0 | 0 | 217.5 | 100% | 11322 |
| Q8_0 | 4k | 204.5 | 94% | 10335 |
| Q8_0 | 16k | 201.7 | 93% | 9258 |
Alternatives
Less memory that still fits Gemma 4 26B-A4B
Smaller memory usually means a cheaper card or Mac (we do not list prices).
- Mac mini M6 32GB: UD-Q5_K_M, 22 GB usable, ~27 tok/s estimated
- MacBook Air M5 32GB: UD-Q5_K_M, 22 GB usable, ~25 tok/s estimated
- AMD RX 7900 XTX 24GB: UD-Q5_K_M, 23 GB usable, ~148 tok/s estimated
Other models on RTX 5090 32GB, similar size
- Mistral Small 3.2 24B: Q8_0, fits fully, ~64 tok/s (measured)
- Qwen3.6 27B: Q8_0, fits fully, ~52 tok/s (measured)
- Qwen3.8 27B (hardware page): Q8_0, fits fully, ~53 tok/s (scaled from measured)
- LFM2.5 24B-A2B (hardware page): Q8_0, fits fully, ~343 tok/s (scaled from measured)
Memory and context limits
Q8_0 weights are 26.9 GB; the 16k f16 KV cache adds 0.55 GB. Largest context that fits in 31.0 GB usable:
| KV cache type | Max context |
|---|---|
| f16 | 128k |
| q8_0 | 256k |
| q4_0 | 256k |
Native limit 256k. Full KV cache table for Gemma 4 26B-A4B.
Copy-paste commands
The advisor's pick for this hardware: Q8_0, 16k context.
llama-server -hf unsloth/gemma-4-26B-A4B-it-GGUF:Q8_0 -c 16384 -fa on --jinja -ngl 99
OLLAMA_CONTEXT_LENGTH=16384 ollama serve ollama run hf.co/unsloth/gemma-4-26B-A4B-it-GGUF:Q8_0
FAQ
How fast is Gemma 4 26B-A4B on RTX 5090 32GB?
About 217.5 tokens/s for generation (Q8_0 on llama.cpp CUDA, 128-token generation), 11322 tokens/s for prompt processing. Speed drops as the context fills; see the depth table.
How much memory does Gemma 4 26B-A4B need on RTX 5090 32GB?
The advisor's pick, Q8_0, is 26.9 GB of weights plus 0.55 GB of KV cache at 16k context (27.4 GB), against 31.0 GB usable.
Is the Gemma 4 26B-A4B speed on RTX 5090 32GB measured or estimated?
Measured: our own llama-bench run on a rented GeForce RTX 5090, including runs with 4k to 16k tokens already in context.
What can I use instead of RTX 5090 32GB for Gemma 4 26B-A4B?
With less memory, Mac mini M6 32GB (~27 tok/s), MacBook Air M5 32GB (~25 tok/s), AMD RX 7900 XTX 24GB (~148 tok/s) also fit it fully.
What is the maximum context for Gemma 4 26B-A4B on RTX 5090 32GB?
With an f16 KV cache about 128k tokens, 256k with q8_0 and 256k with q4_0, bounded by the model's own limit.
Related
All models on RTX 5090 32GB · Gemma 4 26B-A4B on all hardware · VRAM tier · all measured pages · leaderboard
Compare: RTX 4090 24GB vs RTX 5090 32GB · RTX 3090 24GB vs RTX 5090 32GB · 2x RTX 3090 (48GB total) vs RTX 5090 32GB