Qwen 0.8B AgentJSON — Hardware Requirements & GPU Compatibility
ChatFunctions
Qwen 0.8B AgentJSON is a 752M-parameter open language model from airev-ae in the Qwen family. It supports a context window of up to 262,144 tokens. At BF16 it needs about 1.86 GB of VRAM — see which GPUs and Macs can run it below.
Runs great on a 24 GB GPU
Runs from 8 GB (BF16 · 256K)Best on 24 GB (BF16 · ~352 tok/s · 256K)Macs from 8 GB (BF16 · ~39 tok/s)
Apache 2.0
Reference GPU · RTX 4090 24 GB
Estimates, ±20% · how we estimate
Run Qwen 0.8B AgentJSON
One recommended way. Every other method, OS and quant is in the step-by-step guide.Each way opens a step-by-step guide for your OS and quant.
Recommended
vllm serve airev-ae/Qwen-0.8B-AgentJSON --max-model-len 16384vLLM repo with its own quantisation
Download: Weights on Hugging Face
Ways to runOther ways
each opens step-by-stepRun in cloud
Runs on RTX 3080 Ti (12 GB) (10 GB headroom) · BF16
- Generation speed
- ~319 tok/s
- generation speed
- Cost per 1M output tokens
- $0.11
- per 1M output tokens
How Much VRAM Does Qwen 0.8B AgentJSON Need?
Select a quantization level; the listed hardware adheres to it.
| Quantization | VRAM | +32K | On RTX 4090 (reference) | |
|---|---|---|---|---|
| 1.9 GB | 2 GB | 16-bit · bfloat — Brain floating point 16 — preferred for training | Fits: 256K context · ~352 tok/s |
Selected BF16
Brain floating point 161.9GBat 16K context
- Weights
- 1.5 GB
- KV cache
- 0.1 GB
- Overhead
- 0.3 GB
Fits RTX 4090 (reference) · 22.1 GB free · ~352 tok/s
≥50 tok/s needs ≥143 GB/s · RTX 5090 ~626, RTX 3090 Ti ~352
Which Hardware Can Run Qwen 0.8B AgentJSON?
BF16 · 1.9 GBQwen 0.8B AgentJSON (BF16) requires 1.9 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 3+ GB is recommended. Using the full 262K context window can add up to 1.6 GB, bringing total usage to 3.4 GB. 52 GPUs can run it, including NVIDIA GeForce RTX 5090, NVIDIA GeForce RTX 3090 Ti. 163 devices with unified memory can run Qwen 0.8B AgentJSON, including Mac Studio (M3 Ultra, 256GB), Mac Studio (M3 Ultra, 512GB).
≤ 11 GB
≤ 11 GB tier →fits · 6.1 GB free · up to 256K ctx
16 GPUs · 1 Mac class · 2 device families
RTX 3080~266Simulate output on RTX 3080RTX 2080 Ti~215Simulate output on RTX 2080 TiRTX 3070 Ti~213Simulate output on RTX 3070 TiM3*8 GB~39Simulate output on iMac 24" M3 (8 GB)Orin Nano~36Simulate output on NVIDIA Jetson Orin Nano 8GB (Super)iPhone~26Simulate output on iPhone 1720 GB
fits · 18.1 GB free · up to 256K ctx
1 GPU
RX 7900 XT~258Simulate output on RX 7900 XT32–40 GB
fits · 30.1 GB free · up to 256K ctx
1 GPU · 3 Mac classes · 4 device families
RTX 5090~626Simulate output on RTX 5090M5 Max36 GB~173Simulate output on Mac Studio M5 Max (36 GB)M532 GB~58Simulate output on MacBook Air 13" M5 (32 GB)Orin~72Simulate output on NVIDIA Jetson AGX Orin 32GBSnapdragon X Elite (X1E-80-100)~47Simulate output on Snapdragon X Elite Copilot+ PC
By VRAM tier · typical GPUs and Macs at BF16
Alternatives to Qwen 0.8B AgentJSON
Qwen family · 28 models →Similar models
4| Model | Score | VRAM | tok/s | Why it is similar |
|---|---|---|---|---|
| Qwen3 0.6BAlibaba | — | 1.7 GB | ~753 | same size |
| Sarashina2.2 0.5B Instruct v0.1sbintuitions | — | 1.3 GB | ~728 | same size |
| Gpt2 LargeOpenAI | — | 0.5 GB | ~1213 | same size |
| OneReason 0.8B Pretrain CompetitionOpenOneRec | — | 1.7 GB | ~728 | same size |
Specifications
- Publisher
- airev-ae
- Family
- Qwen
- Parameters
- 752M
- Architecture
- Qwen3_5ForCausalLM
- Context Length
- 262,144 tokens
- Vocabulary Size
- 248,320
- Release Date
- 2026-03-25
- License
- Apache 2.0
Frequently Asked Questions
How much VRAM does Qwen 0.8B AgentJSON need?
Qwen 0.8B AgentJSON requires 1.9 GB of VRAM at BF16. Full 262K context adds up to 1.6 GB (3.4 GB total).
VRAM = Weights + KV Cache + Overhead
Weights = 752M × 16 bits ÷ 8 = 1.5 GB
KV Cache + Overhead ≈ 0.4 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 ≈ 1.9 GB (at full 262K context)
VRAM usage by quantization
Can I run Qwen 0.8B AgentJSON on a Mac?
Yes — MacBook Air 13" M3 (8 GB) and 145 other Macs can run Qwen 0.8B AgentJSON. Apple Silicon uses unified memory, so the model shares RAM with the system. At BF16 you need at least 1.9 GB of usable unified memory (RAM minus macOS overhead).
Can I run Qwen 0.8B AgentJSON locally?
Yes — Qwen 0.8B AgentJSON can run locally on consumer hardware. Popular tools include Ollama, LM Studio, and llama.cpp.
How fast is Qwen 0.8B AgentJSON?
At BF16, Qwen 0.8B AgentJSON can reach ~2581 tok/s on AMD Instinct MI350X. On NVIDIA GeForce RTX 4090: ~352 tok/s. Speed depends mainly on GPU memory bandwidth. Real-world results typically within ±20%.
tok/s = (bandwidth GB/s ÷ model GB) × efficiency
Example: NVIDIA B200 → 8000 ÷ 1.86 × 0.65 = ~2796 tok/s
Estimated speed at BF16 (1.9 GB)
Real-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.
What's the download size of Qwen 0.8B AgentJSON?
At BF16, the download is about 1.50 GB.
Which GPUs can run Qwen 0.8B AgentJSON?
52 consumer GPUs can run Qwen 0.8B AgentJSON at BF16. Top options include AMD Radeon RX 6700 XT, AMD Radeon RX 6800, AMD Radeon RX 6800 XT. 52 GPUs have plenty of headroom for comfortable inference.
Which devices can run Qwen 0.8B AgentJSON?
163 devices with unified memory can run Qwen 0.8B AgentJSON at BF16 (1.9 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.