Can Qwen 3.5 9B run on RTX 5070 12GB?

YES — Runs Great

S98Excellent

Estimated from fit model

Qwen 3.5 9B needs ~9.8 GB VRAM. RTX 5070 12GB has 12.0 GB. With Q4_K_M quantization, expect ~83 tok/s.

Runtime: llama.cppCapacity: RoomyBandwidth: MediumStack: StandardBottleneck: 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.

Capabilities:

Select quantization to explore

Q4_K_M (Medium quality) — 9.8 GB, 82.9 tok/s, Runs well

9.8 GB required12.0 GB available

82% VRAM used

Fit status

Runs well

Decode

82.9 tok/s

TTFT

2335 ms

Safe context

32K

Memory

9.8 GB / 12.0 GB

Memory breakdown

Weights5.5 GB

KV Cache2.2 GB

Runtime0.9 GB

Headroom1.2 GB

See how fast it feels

See how fast it feelsQwen 3.5 9B on RTX 5070 12GB
1st promptCold start — includes initialization
>Write a Python function to sort a list
Here's a Python function that sorts a list using quicksort:

def quicksort(arr):
    if len(arr) <= 1:
        return arr
    pivot = arr[len(arr) // 2]
    left   = [x for x in arr if x < pivot]
    middle = [x for x in arr if x == pivot]
    right  = [x for x in arr if x > pivot]
    return quicksort(left) + middle + quicksort(right)

# Example usage
numbers = [3, 6, 8, 10, 1, 2, 1]
print(quicksort(numbers))  # [1, 1, 2, 3, 6, 8, 10]

The algorithm picks a pivot, partitions the list into smaller, equal, and larger elements, then recursively sorts each partition and concatenates the results.
2nd promptWarm — model ready, no init overhead
>Now add type hints and a docstring
def quicksort(arr: list[int]) -> list[int]:
    """Sort a list using the quicksort algorithm.

    Args:
        arr: List of integers to sort.

    Returns:
        A new sorted list.
    """
    if len(arr) <= 1:
        return arr
    pivot = arr[len(arr) // 2]
    left   = [x for x in arr if x < pivot]
    middle = [x for x in arr if x == pivot]
    right  = [x for x in arr if x > pivot]
    return quicksort(left) + middle + quicksort(right)

Added type hints for the input and return type, plus a Google-style docstring.
Estimated: 82.9 tok/s decode · 2.3s TTFT (warm) · 207 tok/s prefill

What limits this setup

This setup is broadly balanced for this model.

No major red flags

This recommendation has enough memory headroom and acceptable estimated speed for the selected workload.

Best improvement path

Performance by workload

Workload	Grade	Fit	Decode	TTFT	Context
Chat	S	Runs well	82.9 tok/s	1274 ms	32K
Coding	S	Runs well	82.9 tok/s	2335 ms	32K
Agentic Coding	S	Runs with offload	82.9 tok/s	3397 ms	32K
Reasoning	S	Runs well	82.9 tok/s	2760 ms	32K
RAG	S	Runs with offload	82.9 tok/s	4246 ms	32K

Inference speed

Qwen 3.5 9B inference speed — tokens per second by GPU & Mac

Estimated decode speed (tokens/sec) for Qwen 3.5 9B at Q4_K_M across popular GPUs and Apple Silicon, using the fastest local runtime per device. Fastest is RTX 5090 32GB at ~126 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	126.0	Fits
RTX 4090 24GB	24 GB	Q4_K_M	126.0	Fits
RTX 3090 24GB	24 GB	Q4_K_M	126.0	Fits
RX 7900 XTX 24GB	24 GB	Q4_K_M	126.0	Fits
RTX 4080 Super 16GB	16 GB	Q4_K_M	119.6	Fits
Mac Studio M3 Ultra 256GB	256 GB	Q4_K_M	109.1	Fits
Mac Studio M2 Ultra 128GB	128 GB	Q4_K_M	90.9	Fits
Mac Studio M1 Ultra 128GB	128 GB	Q4_K_M	86.2	Fits
MacBook Pro M4 Max 128GB	128 GB	Q4_K_M	73.4	Fits
MacBook Pro M4 Max 64GB	64 GB	Q4_K_M	73.4	Fits
RTX 4070 12GB	12 GB	Q4_K_M	71.5	Fits
MacBook Pro M3 Max 64GB	64 GB	Q4_K_M	47.0	Fits
MacBook Pro M1 Max 64GB	64 GB	Q4_K_M	43.1	Fits
RTX 3060 12GB	12 GB	Q4_K_M	39.9	Fits
MacBook Pro M4 Pro 48GB	48 GB	Q4_K_M	37.9	Fits
RTX 4060 8GB	8 GB	Q4_K_M	19.2	Heavy offload

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 Qwen 3.5 9B (9B params) fits at each quantization level on RTX 5070 12GB (12.0 GB usable).

Quant	Bits	VRAM	Quality	Fit
Q2_K	2	3.5 GB	Low	S92
Q3_K_S	3	4.4 GB	Low	S93
NVFP4	4	5.0 GB	Medium	S94
Q4_K_M	4	5.5 GB	Medium	S94
Q5_K_M	5	6.5 GB	High	S94
Q6_KBest for your GPU	6	7.4 GB	High	S93
Q8_0	8	9.6 GB	Very High	F0
F16	16	18.5 GB	Maximum	F0

Get started

Copy-paste commands to run Qwen 3.5 9B on your machine.

Run

ollama run qwen3.5:9b

Frequently asked questions

Can RTX 5070 12GB run Qwen 3.5 9B?

Yes, RTX 5070 12GB can run Qwen 3.5 9B with a S grade (Runs well). Expected decode speed: 82.9 tok/s.

How much VRAM does Qwen 3.5 9B need?

Qwen 3.5 9B (9B parameters) requires approximately 9.8 GB of memory with Q4_K_M quantization.

What is the best quantization for Qwen 3.5 9B?

The recommended quantization for Qwen 3.5 9B is Q4_K_M, which balances quality and memory efficiency.

What speed will Qwen 3.5 9B run at on RTX 5070 12GB?

On RTX 5070 12GB, Qwen 3.5 9B achieves approximately 82.9 tokens per second decode speed with a time-to-first-token of 2335ms using Q4_K_M quantization.

Can RTX 5070 12GB run Qwen 3.5 9B for coding?

For coding workloads, Qwen 3.5 9B on RTX 5070 12GB receives a S grade with 82.9 tok/s and 32K context.

What context window can Qwen 3.5 9B use on RTX 5070 12GB?

On RTX 5070 12GB, Qwen 3.5 9B can safely use up to 32K tokens of context. The model's official context limit is 131K, but available memory constrains the safe maximum.

See all results for RTX 5070 12GB See all hardware for Qwen 3.5 9B

Embed this result▼

Paste this snippet into any page to show a live fit card.

<iframe src="https://willitrunai.com/embed/qwen-3.5-9b-on-rtx-5070-12gb" width="400" height="180" frameborder="0" style="border:none;border-radius:12px;overflow:hidden;" title="Will It Run AI — fit result"></iframe>

Preview: