Qwen3.8 Flash Coder 85gb BF16 — Hardware Requirements & GPU Compatibility
ChatCode
Qwen3.8 Flash Coder 85gb BF16 is a 42.6B-parameter open language model from Jab1718 in the Qwen 3.8 family. It supports a context window of up to 262,144 tokens. At BF16 it needs about 85.62 GB of VRAM — see which GPUs and Macs can run it below.
Qwen3.8 Flash Coder 85gb BF16 runs on a 96 GB+ Mac or a workstation GPU
Best local pick → Mac Studio (M3 Ultra, 96GB) · BF16 · 86.2 GB · ~29 tok/s
Macs from 96 GB (BF16 · ~29 tok/s)
Apache 2.0
Not set up for it?
Reference GPU · RTX 4090 24 GB
Estimates, ±20% · how we estimate
Run Qwen3.8 Flash Coder 85gb BF16
One recommended way. Every other method, OS and quant is in the step-by-step guide.
Recommended
vllm serve Jab1718/qwen3.8-flash-coder-85gb-bf16 --max-model-len 16384 --tensor-parallel-size 8vLLM repo with its own quantisation
Download: Weights on Hugging Face
Other ways
each opens step-by-stepRun in cloud
Fits on H200 (141 GB) (54 GB headroom) · BF16
- Generation speed
- ~137 tok/s
- generation speed
- Cost per 1M output tokens
- $7.27
- per 1M output tokens
How Much VRAM Does Qwen3.8 Flash Coder 85gb BF16 Need?
Select a quantization level; the listed hardware adheres to it.
| Quantization | VRAM | +32K | Runs on | |
|---|---|---|---|---|
| 85.6 GB | 86.9 GB | 16-bit · bfloat — Brain floating point 16 — preferred for training | Runs on RTX PRO 6000 Blackwell Max-Q Workstation Edition 96 GB |
Selected BF16
Brain floating point 1686.2GBat 16K context
- Weights
- 85.2 GB
- KV cache
- 0.7 GB
- Overhead
- 0.3 GB
Server-class: ~86.2 GB at 16K context · one RTX PRO 6000 Blackwell Max-Q Workstation Edition 96 GB
Which Hardware Can Run Qwen3.8 Flash Coder 85gb BF16?
BF16 · 85.6 GBQwen3.8 Flash Coder 85gb BF16 (BF16) requires 85.6 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 112+ GB is recommended. Using the full 262K context window can add up to 10.7 GB, bringing total usage to 96.3 GB. No consumer GPU has enough memory. 30 devices with unified memory can run Qwen3.8 Flash Coder 85gb BF16, including Mac Studio (M3 Ultra, 256GB), Mac Studio (M3 Ultra, 512GB), Mac Studio (M3 Ultra, 96GB).
96 GB
fits · 9.8 GB free · up to 128K ctx
3 GPUs · 2 Mac classes
RTX PRO 6000 Blackwell Max-Q Workstation Edition~66Simulate output on RTX PRO 6000 Blackwell Max-Q Workstation EditionRTX PRO 6000 Blackwell Workstation Edition~66Simulate output on RTX PRO 6000 Blackwell Workstation EditionRTX PRO 6000 Blackwell Server Edition~60Simulate output on RTX PRO 6000 Blackwell Server EditionM3 Ultra*96 GB~29Simulate output on Mac Studio (M3 Ultra, 96GB)M3 Max*96 GB~13Simulate output on MacBook Pro 14" M3 Max (96 GB)128–144 GB
fits · 41.8 GB free · up to 256K ctx
4 GPUs · 2 Mac classes · 4 device families
GH200 Grace Hopper Superchip~139Simulate output on GH200 Grace Hopper SuperchipH200 NVL~137Simulate output on H200 NVLH200 SXM~137Simulate output on H200 SXMM2 Ultra*128 GB~28Simulate output on Mac Pro M2 Ultra (128 GB)M5 Max128 GB~23Simulate output on Mac Studio M5 Max (128 GB)GB10~12Simulate output on NVIDIA DGX SparkJetson T5000 (Thor)~12Simulate output on NVIDIA Jetson AGX Thor Developer Kit
By VRAM tier · typical GPUs and Macs at BF16
Consumer GPUs (up to 32 GB) and Macs up to 64 GB: too small for this model (needs ≥ 86.2 GB)8 GB tier →12 GB tier →16 GB tier →24 GB tier →Macs →
Alternatives to Qwen3.8 Flash Coder 85gb BF16
Qwen 3.8 family · 12 models →Similar models
3| Model | Score | VRAM | tok/s | Why it is similar |
|---|---|---|---|---|
| Phi 3.5 MoE InstructMicrosoft | — | 23.9 GB | ~141 | same size |
| Falcon 40B InstructTII UAE | — | 23.8 GB | ~33 | same size |
| IQuest Coder V1 40B Loop InstructIQuestLab | — | 85.3 GB | —needs > 24 GB | same size |
Specifications
- Publisher
- Jab1718
- Family
- Qwen 3.8
- Parameters
- 42.6B
- Active parameters
- 7.2B per token
- Architecture
- Qwen4ExpForCausalLM
- Context Length
- 262,144 tokens
- Vocabulary Size
- 248,320
- Release Date
- 2026-08-27
- License
- Apache 2.0
Frequently Asked Questions
How much VRAM does Qwen3.8 Flash Coder 85gb BF16 need?
Qwen3.8 Flash Coder 85gb BF16 requires 85.6 GB of VRAM at BF16. Full 262K context adds up to 10.7 GB (96.3 GB total).
VRAM = Weights + KV Cache + Overhead
Weights = 42.6B × 16 bits ÷ 8 = 85.2 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 ≈ 11.1 GB (at full 262K context)
VRAM usage by quantization
Can NVIDIA GeForce RTX 5090 run Qwen3.8 Flash Coder 85gb BF16?
No — Qwen3.8 Flash Coder 85gb BF16 requires at least 85.6 GB at BF16, which exceeds the NVIDIA GeForce RTX 5090's 32 GB of VRAM.
Can I run Qwen3.8 Flash Coder 85gb BF16 on a Mac?
Yes — Mac Studio (M3 Ultra, 96GB) and 20 other Macs can run Qwen3.8 Flash Coder 85gb BF16. Apple Silicon uses unified memory, so the model shares RAM with the system. At BF16 you need at least 85.6 GB of usable unified memory (RAM minus macOS overhead).
Can I run Qwen3.8 Flash Coder 85gb BF16 locally?
Yes — Qwen3.8 Flash Coder 85gb BF16 can run locally on consumer hardware. Popular tools include Ollama, LM Studio, and llama.cpp.
How fast is Qwen3.8 Flash Coder 85gb BF16?
At BF16, Qwen3.8 Flash Coder 85gb BF16 can reach ~79 tok/s on AMD Instinct MI350X. 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 → 14.5 GB active ÷ (8000 × 0.65) = 2.79 ms, plus 48 layers × 0.055 ms = 2.64 ms, so 1000 ÷ 5.43 ms = ~184 tok/s
Estimated speed at BF16 (85.6 GB)
Real-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.
What's the download size of Qwen3.8 Flash Coder 85gb BF16?
At BF16, the download is about 85.24 GB.
Which GPUs can run Qwen3.8 Flash Coder 85gb BF16?
No single consumer GPU has enough VRAM to run Qwen3.8 Flash Coder 85gb BF16 at BF16 (85.6 GB). Multi-GPU or professional hardware is required.
Which devices can run Qwen3.8 Flash Coder 85gb BF16?
30 devices with unified memory can run Qwen3.8 Flash Coder 85gb BF16 at BF16 (85.6 GB), including ASUS Ascent GX10, Beelink GTR9 Pro (Ryzen AI Max+ 395, 128 GB), Framework Desktop (Ryzen AI Max+ 395, 128 GB), GMKtec EVO-X2 (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.