RTX PRO 5000 Blackwell 48GBGünstige Wahl48 GB VRAM (+16)
C
Adds memory headroom for longer context windows and future model growth.52.9 tok/s Dekodierung
Adds memory headroom for longer context windows and future model growth.
ca. $4,999 MSRP
Can Qwen3.5 35B A3B run on RTX 5090 32GB?
YES — Tight Fit
C53Usable
Estimated from fit model
Qwen3.5 35B A3B needs ~29.9 GB VRAM. RTX 5090 32GB has 32.0 GB. With Q4_K_M quantization, expect ~56 tok/s.
Runtime: OllamaCapacity: TightBandwidth: HighStack: BasicBottleneck: Balanced
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) — 29.9 GB, 56.2 tok/s, Tight fit
29.9 GB required32.0 GB available
93% VRAM used
Fit status
Tight fit
Decode
56.2 tok/s
TTFT
3443 ms
Safe context
24K
Memory
29.9 GB / 32.0 GB
Memory breakdown
Weights21.3 GB
KV Cache4.1 GB
Runtime1.2 GB
Headroom3.2 GB
See how fast it feels
See how fast it feelsQwen3.5 35B A3B on RTX 5090 32GB
1st promptCold start — includes initialization
>What is local AI inference?
Local AI inference means running an AI language model directly on your own hardware — your laptop, desktop, or server — instead of sending requests to a remote cloud API.
When you run inference locally the model weights are loaded into your GPU or unified memory. Each token you generate requires reading those weights from memory, so memory bandwidth is the main bottleneck for decode speed.
Key benefits of running locally:
- Full privacy: your prompts never leave your machine
- No per-token cost or rate limits
- Works offline once the model is downloaded
- Latency depends only on your hardware
2nd promptWarm — model ready, no init overhead
>How much VRAM do I need?
It depends on the model size and quantization level. A rough rule of thumb:
Model size Q4 (4-bit) Q8 (8-bit) FP16
7B params ~4.3 GB ~7.5 GB ~14 GB
13B params ~7.9 GB ~13.9 GB ~26 GB
70B params ~42.7 GB ~74.9 GB ~140 GB
Most people use 4-bit quantization (Q4_K_M) which gives 90-95% of full quality at a fraction of the memory. A 24 GB GPU can comfortably run most 7B-13B models.
Estimated: 56.2 tok/s decode · 3.4s TTFT (warm) · 141 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 | C | Tight fit | 56.2 tok/s | 1878 ms | 24K |
| Coding | C | Tight fit | 56.2 tok/s | 3443 ms | 24K |
| Agentic Coding | C | Runs with offload (needs ~1.2 GB host RAM) | 38.1 tok/s | 7389 ms | 24K |
| Reasoning | C | Tight fit | 56.2 tok/s | 4069 ms | 24K |
| RAG | C | Runs with offload (needs ~1.2 GB host RAM) | 38.1 tok/s | 9237 ms | 24K |
Inference speed
Qwen3.5 35B A3B inference speed — tokens per second by GPU & Mac
Estimated decode speed (tokens/sec) for Qwen3.5 35B A3B at Q4_K_M across popular GPUs and Apple Silicon, using the fastest local runtime per device. Fastest is RTX 5090 32GB at ~56 tok/s. Speed is memory-bandwidth bound, so cards that fit the whole model in VRAM run far faster than ones that offload to system RAM.
| GPU / Mac | Memory | Quant | Speed (tok/s) | Fits? |
|---|---|---|---|---|
| RTX 5090 32GB | 32 GB | Q4_K_M | 56.2 | Tight |
MacBook Pro M4 Max 128GB | 128 GB | Q4_K_M | 28.1 | Fits |
MacBook Pro M4 Max 64GB | 64 GB | Q4_K_M | 28.1 | Fits |
Mac Studio M3 Ultra 256GB | 256 GB | Q4_K_M | 26.1 | Fits |
Mac Studio M2 Ultra 128GB | 128 GB | Q4_K_M | 21.7 | Fits |
Mac Studio M1 Ultra 128GB | 128 GB | Q4_K_M | 20.6 | Fits |
MacBook Pro M4 Pro 48GB | 48 GB | Q4_K_M | 17.7 | Tight |
RX 7900 XTX 24GB | 24 GB | Q4_K_M | 16.6 | Heavy offload |
MacBook Pro M3 Max 64GB | 64 GB | Q4_K_M | 11.2 | Fits |
| RTX 4090 24GB | 24 GB | Q4_K_M | 10.6 | Heavy offload |
MacBook Pro M1 Max 64GB | 64 GB | Q4_K_M | 10.3 | Fits |
| RTX 3090 24GB | 24 GB | Q4_K_M | 9.7 | Heavy offload |
| RTX 4080 Super 16GB | 16 GB | Q4_K_M | 6.5 | Too big |
| RTX 4070 12GB | 12 GB | Q4_K_M | 2.7 | Too big |
| RTX 3060 12GB | 12 GB | Q4_K_M | 2.0 | Too big |
| RTX 4060 8GB | 8 GB | Q4_K_M | 2.0 | Too big |
Estimates for single-stream decoding at Q4_K_M; real tokens/sec varies with prompt length, context, batch size, and runtime build. Prompt processing (prefill) is faster than the decode figures shown here.
Quantization options
How Qwen3.5 35B A3B (35B params) fits at each quantization level on RTX 5090 32GB (32.0 GB usable).
| Quant | Bits | VRAM | Quality | Fit |
|---|---|---|---|---|
Q2_K | 2 | 13.7 GB | Low | C49 |
Q3_K_S | 3 | 17.2 GB | Low | C50 |
NVFP4 | 4 | 19.6 GB | Medium | C49 |
Q4_K_M | 4 | 21.3 GB | Medium | C49 |
Q5_K_MBest for your GPU | 5 | 25.2 GB | High | C49 |
Q6_K | 6 | 28.7 GB | High | F0 |
Q8_0 | 8 | 37.5 GB | Very High | F0 |
F16 | 16 | 71.8 GB | Maximum | F0 |
Get started
Copy-paste commands to run Qwen3.5 35B A3B on your machine.
Run
docker run --rm -it ghcr.io/ggerganov/llama.cpp:full \
--hf-repo "unsloth/Qwen3.5-35B-A3B-GGUF" \
--hf-file "Qwen3.5-35B-A3B-GGUF-Q4_K_M.gguf" \
-c 4096 -ngl 99Upgrade-Optionen
Hardware, die Qwen3.5 35B A3B gut ausführt
RTX PRO 5000 Blackwell 48GBGünstige WahlNVIDIA L40 48GBBestes Preis-Leistungs-Verhältnis48 GB VRAM (+16)
C
Adds memory headroom for longer context windows and future model growth.31.6 tok/s Dekodierung
Adds memory headroom for longer context windows and future model growth.
ca. $5,500 MSRP
RTX 6000 Ada 48GBNVIDIA-Upgrade48 GB VRAM (+16)
C
Adds memory headroom for longer context windows and future model growth.36.9 tok/s Dekodierung
Adds memory headroom for longer context windows and future model growth.
ca. $6,800 MSRP
Frequently asked questions
Can RTX 5090 32GB run Qwen3.5 35B A3B?
Yes, RTX 5090 32GB can run Qwen3.5 35B A3B with a C grade (Tight fit). Expected decode speed: 56.2 tok/s.
How much VRAM does Qwen3.5 35B A3B need?
Qwen3.5 35B A3B (35B parameters) requires approximately 29.9 GB of memory with Q4_K_M quantization.
What is the best quantization for Qwen3.5 35B A3B?
The recommended quantization for Qwen3.5 35B A3B is Q4_K_M, which balances quality and memory efficiency.
What speed will Qwen3.5 35B A3B run at on RTX 5090 32GB?
On RTX 5090 32GB, Qwen3.5 35B A3B achieves approximately 56.2 tokens per second decode speed with a time-to-first-token of 3443ms using Q4_K_M quantization.
Can RTX 5090 32GB run Qwen3.5 35B A3B for coding?
For coding workloads, Qwen3.5 35B A3B on RTX 5090 32GB receives a C grade with 56.2 tok/s and 24K context.
What context window can Qwen3.5 35B A3B use on RTX 5090 32GB?
On RTX 5090 32GB, Qwen3.5 35B A3B can safely use up to 24K tokens of context. The model's official context limit is —, but available memory constrains the safe maximum.
What should I upgrade first if Qwen3.5 35B A3B feels slow on RTX 5090 32GB?
Buy headroom, not only minimum fit. A slightly larger memory tier gives you safer context growth and makes the recommendation more future-proof.
Embed this result▼
Paste this snippet into any page to show a live fit card.
<iframe src="https://willitrunai.com/embed/hf-unsloth--qwen3-5-35b-a3b-gguf-on-rtx-5090-32gb" width="400" height="180" frameborder="0" style="border:none;border-radius:12px;overflow:hidden;" title="Will It Run AI — fit result"></iframe>
Preview: