llmrun
hotdogsQwenQwen3_5MoeForConditionalGeneration

Qwen35 Agent R2A103 — Hardware Requirements & GPU Compatibility

ChatFunctionsReasoning

Qwen35 Agent R2A103 is a 35.1B-parameter open language model from hotdogs in the Qwen family. It supports a context window of up to 262,144 tokens. At Q4_K_M it needs about 21.45 GB of VRAM — see which GPUs and Macs can run it below.

35.1B params · MoE3.9B active262K contextApache 2.0621 downloads2 likes
Based on Qwen35B Agent R2
A

Runs well on a 24 GB GPU

Runs from16 GB Q2_K · 64KBest on24 GB Q4_K_M · ~171 tok/s · 128KMacs from24 GB Q3_K_M · ~59 tok/s

Apache 2.0

Reference GPU · RTX 4090 24 GB

Estimates, ±20% · how we estimate

Run Qwen35 Agent R2A103

One recommended way. Every other method, OS and quant is in the step-by-step guide.

Recommended

vLLM
one command · needs vLLM installed
$vllm serve hotdogs/Qwen35-Agent-R2A103 --max-model-len 16384

vLLM repo with its own quantisation

Download: Weights on Hugging Face

install, OS, other quants

Other ways

each opens step-by-step

Run in cloud

Fits on RTX A6000 (48 GB) (26 GB headroom) · Q4_K_M

Generation speed
~143 tok/s
generation speed
Cost per 1M output tokens
$0.64
per 1M output tokens
Deployon RunPodBest optionfrom $0.33/hr (opens in new tab)
Compare GPUs →
or

How Much VRAM Does Qwen35 Agent R2A103 Need?

Select a quantization level; the listed hardware adheres to it.

QuantizationVRAM+32KOn RTX 4090 (reference)
15.3 GB15.6 GB2-bit · K-quant — 2-bit quantization with K-quant improvementsFits: 256K context · ~209 tok/s
17.5 GB17.8 GB3-bit medium — 3-bit medium quantizationFits: 256K context · ~194 tok/s
21.4 GB21.7 GB4-bit medium — 4-bit medium quantization — most popular sweet spotFits: 128K context · ~171 tok/s
25.4 GB25.7 GB5-bit medium — 5-bit medium quantization — good quality/size tradeoffDoes not fit: needs 25.5 GB
29.4 GB29.6 GB6-bit, very good — 6-bit quantization, very good qualityDoes not fit: needs 29.4 GB
35.5 GB35.7 GB8-bit, near-lossless — 8-bit quantization, near-losslessDoes not fit: needs 35.6 GB
70.6 GB70.9 GB16-bit · bfloat — Brain floating point 16 — preferred for trainingDoes not fit: needs 70.7 GB
est.calculated, no published file yet

Selected Q4_K_M

4-bit medium quantization

21.5GBat 16K context

Weights
21.1 GB
KV cache
0.2 GB
Overhead
0.3 GB

Fits RTX 4090 (reference) · 2.5 GB free · ~171 tok/s

≥50 tok/s needs ≥206 GB/s · RTX 5090 ~236, RTX 3090 Ti ~171

Or rent RTX A6000 from $0.33/hr.

Which Hardware Can Run Qwen35 Agent R2A103?

Q4_K_M · 21.4 GB

Qwen35 Agent R2A103 (Q4_K_M) requires 21.4 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 28+ GB is recommended. Using the full 262K context window can add up to 2.6 GB, bringing total usage to 24.1 GB. 7 GPUs can run it, including NVIDIA GeForce RTX 5090. 100 devices with unified memory can run Qwen35 Agent R2A103, including Mac Studio (M3 Ultra, 256GB), Mac Studio (M3 Ultra, 512GB), Mac Studio M1 Max (32 GB).

By VRAM tier · typical GPUs and Macs at Q4_K_M
8–12 GB · RTX 4060, RTX 3060 12GBDoesn't fit — smallest quant needs 15.4 GB8 GB tier →12 GB tier →
16 GB · RTX 5060 Ti 16GBQ2_K · 15.4 GB · ~124 tok/s · 64K16 GB tier →
24 GB · RTX 4090referenceQ4_K_M · 21.5 GB · ~171 tok/s · 128K24 GB tier →
Mac 32 GB · Mac Mini M4Q4_K_M · 21.5 GB · ~28 tok/s · 256KMacs →
Mac 64 GB · Mac Studio M4 MaxQ4_K_M · 21.5 GB · ~70 tok/s · 256KMacs →

Alternatives to Qwen35 Agent R2A103

Qwen family · 28 models →

Similar models

6
ModelScoreVRAMtok/sWhy it is similar
MiniCPM V 4.7 35B A3BOpenBMB—22.1 GB~169same size
Llm Jp 4.1 33B Thinkingllm-jp—22.5 GB~35same size
Agnes 3.0 FlashAgnes-AI—22.1 GB~35same size
Ornith 1.5 35B A3Bornith-ai—22 GB~151same size
Agents A1InternScience—21.5 GB~171same size
Intern S2 PreviewInternLM—73.2 GB—needs > 24 GBsame size

Specifications

Publisher
hotdogs
Family
Qwen
Parameters
35.1B
Active parameters
3.9B per token
Architecture
Qwen3_5MoeForConditionalGeneration
Context Length
262,144 tokens
Vocabulary Size
248,320
Release Date
2026-07-15
License
Apache 2.0

Resources

Frequently Asked Questions

How much VRAM does Qwen35 Agent R2A103 need?

Qwen35 Agent R2A103 requires 21.4 GB of VRAM at Q4_K_M, or 70.6 GB at BF16. Full 262K context adds up to 2.6 GB (24.1 GB total).

VRAM = Weights + KV Cache + Overhead

Weights = 35.1B × 4.8 bits ÷ 8 = 21.1 GB

KV Cache + Overhead ≈ 0.3 GB (at 2K context + ~0.3 GB framework)

Fit ratings and hardware model lists check this model with room for a 16K-token context, which needs a little more memory.

KV Cache + Overhead ≈ 2.9 GB (at full 262K context)

VRAM usage by quantization

21.4 GB
24.1 GB

Learn more about VRAM estimation →

Can NVIDIA GeForce RTX 4090 run Qwen35 Agent R2A103?

Yes, at Q4_K_M (21.4 GB) or lower. Higher quantizations like Q5_K_M (25.4 GB) exceed the NVIDIA GeForce RTX 4090's 24 GB.

What's the best quantization for Qwen35 Agent R2A103?

For Qwen35 Agent R2A103, Q4_K_M (21.4 GB) offers the best balance of quality and VRAM usage. Q5_K_M (25.4 GB) provides better quality if you have the VRAM. The smallest option is Q2_K at 15.3 GB.

VRAM requirement by quantization

Q2_K
15.3 GB
Q4_K_M ★
21.4 GB
Q5_K_M
25.4 GB
Q6_K
29.4 GB
Q8_0
35.5 GB
BF16
70.6 GB

★ Recommended — best balance of quality and VRAM usage.

Learn more about quantization →

Can I run Qwen35 Agent R2A103 on a Mac?

Yes — Mac Mini M5 Pro (24 GB) and 106 other Macs can run Qwen35 Agent R2A103, the smallest at Q3_K_M. Apple Silicon uses unified memory, so the model shares RAM with the system. At Q4_K_M you need at least 21.4 GB of usable unified memory (RAM minus macOS overhead), from Mac Studio M1 Max (32 GB).

Can I run Qwen35 Agent R2A103 locally?

Yes — Qwen35 Agent R2A103 can run locally on consumer hardware. Popular tools include Ollama, LM Studio, and llama.cpp.

How fast is Qwen35 Agent R2A103?

At Q4_K_M, Qwen35 Agent R2A103 can reach ~118 tok/s on AMD Instinct MI350X. On NVIDIA GeForce RTX 4090: ~171 tok/s. Speed depends mainly on GPU memory bandwidth. Real-world results typically within ±20%.

tok/s = 1000 ÷ (active GB ÷ (bandwidth GB/s × efficiency) × 1000 + layers × routing ms)

Mixture-of-Experts: just the active experts get read for each token, along with a fixed routing cost per layer.

Example: NVIDIA B200 → 2.4 GB active ÷ (8000 × 0.65) = 0.46 ms, plus 40 layers × 0.055 ms = 2.20 ms, so 1000 ÷ 2.66 ms = ~376 tok/s

Estimated speed at Q4_K_M (21.4 GB)

~376 tok/s
~171 tok/s
~376 tok/s
~339 tok/s

Real-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.

Learn more about tok/s estimation →

What's the download size of Qwen35 Agent R2A103?

At Q4_K_M, the download is about 21.06 GB. The full-precision BF16 version is 70.21 GB. The smallest option (Q2_K) is 14.92 GB.

Which GPUs can run Qwen35 Agent R2A103?

7 consumer GPUs can run Qwen35 Agent R2A103 at Q4_K_M. Top options include AMD Radeon RX 7900 XTX, NVIDIA GeForce RTX 3090.

Which devices can run Qwen35 Agent R2A103?

100 devices with unified memory can run Qwen35 Agent R2A103 at Q4_K_M (21.4 GB), including AMD Ryzen AI 9 HX 370 (Strix Point) Laptop, ASUS Ascent GX10, Beelink GTR9 Pro (Ryzen AI Max+ 395, 128 GB), Framework Desktop (Ryzen AI Max+ 395, 128 GB). Apple Silicon Macs use unified memory shared between CPU and GPU, making them well-suited for local LLM inference.