All models

Qwen / LANGUAGE MODEL

Qwen3-235B-A22B-Thinking-2507 VRAM requirements

Explore memory tradeoffs with sourced model data and transparent estimates.

Q4_K_M estimate

152.5 GiB

at 8,192 tokens

FP16 estimate

492.9 GiB

at 8,192 tokens

Suggested GPU

1 × NVIDIA HGX B200 180 GiB

Model architecture

Model computation diagram
  1. Token embedding · 4096
  2. 94 layers · GQA · MoE 8/128 experts active
  3. Output head → token scores
Parameters235,093,634,560 (235.09B)
Layers94
Attention heads64
KV heads4
Head dimension128
Hidden size4096
Maximum context262,144 tokens
NotesStandard transformer KV-cache estimate.

Minimum hardware and family comparison

Estimates use batch 1, an FP16 KV cache and 12% headroom. Multi-GPU capacity assumes even sharding and requires a compatible runtime.

DevicesRequired totalSmallest fitting device exampleUsable per device
1152.51 GiBNVIDIA HGX B200 180 GiB180.0 GiB
2152.85 GiBA100 80 GiB80.0 GiB
4153.52 GiBNVIDIA A100 (40 GiB)40.0 GiB
8154.86 GiBRTX 4000 Ada Generation 20 GiB20.0 GiB

Quantization comparison

Batch 1 · FP16 KV cache · 1 GPU · totals include safety margin. Contexts beyond the model maximum are hypothetical and marked *.

PrecisionWeightsTotal @ 4kTotal @ 8kTotal @ 32k
FP16 / BF1616.0005 bpw · FP16 GGUF reference437.91 GiB492.12 GiB492.94 GiB497.88 GiB
FP88 bpw · Nominal bit width218.95 GiB246.88 GiB247.71 GiB252.64 GiB
INT88 bpw · Nominal bit width218.95 GiB246.88 GiB247.71 GiB252.64 GiB
NVFP66 bpw · Nominal bit width164.21 GiB185.58 GiB186.40 GiB191.34 GiB

GPU architecture and runtime requirements

NVIDIA architectureAttention backendPublished environment or release
Ampere / AdaFlashAttention-2CUDA ≥ 12.0 · PyTorch ≥ 2.2 · FP16 / BF16
Hopper (H100 / H800)FlashAttention-3CUDA ≥ 12.3
Blackwell (B200)FlashAttention-4 · CuTeDSLPyTorch 2.7 introduced Blackwell support in its CUDA 12.8 builds

Transformers version is checkpoint-specific. Model card · Kernel compatibility · PyTorch build notes

Recommended GPUs by memory fit

Q4_K_M @ 8k context, batch 1, FP16 KV cache. Smallest nominal memory first; green results only. Check backend and precision support separately.

Sources & verification

Try your configuration

Your workload

Workload

Precision of the weights held in memory.

Longer contexts make cache precision more important.

Runtime settingsRoofline estimate

Your hardware

Input settings

Sequences per batch. Serving concurrency and throughput require separate measurements.

Advanced runtime settings
Advanced · Model architecture
Layers
94
Attention heads
64
KV heads
4
Head dimension
128
Hidden size
4,096
Maximum context (tokens)
262,144
View model configuration

Performance and memory results

635.5%Memory use

Memory estimate

152.5 GiB needed

Out of memory128.5 GiB over budget24.0 GiB available

Choose more memory, a smaller model, or a lighter workload.

Estimated GPU rental · $0.44/h · Per-GPU quote × devices

This configuration exceeds available memory. Adjust the model or hardware.

Qwen3-235B-A22B-Thinking-2507

RTX 4090 24 GiB

Q4_K_M · FP16 KV

Model specifications
Memory breakdownGiB

Vertical line = usable hardware memory

  • Weights (Q4_K_M)133.95 GiB
  • KV cache (FP16)1.47 GiB
  • Runtime overhead0.75 GiB
  • CUDA Graph capture0.00 GiB · 0.0%
  • Safety margin16.34 GiB
Framework overhead breakdown

Planning allocations. Actual driver, library and allocator use varies by runtime.

Device runtime and driver
0.50 GiB
Operator workspaces
0.20 GiB
CUDA Graph
0.00 GiB
Allocator base
0.05 GiB
Other runtime buffers and communication
0.00 GiB
Compare context lengths
Compare GPU capacity

Inference simulation

Qwen3-235B-A22B-Thinking-2507 · Q4_K_M · RTX 4090 24 GiB

Prompt processing
—
Token generation
—
Estimated duration
—
Timing playback · sample text
Timing playback · sample text0 / 128 · 0 s
This configuration exceeds available memory. Adjust the model or hardware.
Measured latency percentiles and request throughput

Frequently asked questions

How much VRAM does Qwen3-235B-A22B-Thinking-2507 need?

Our default Q4_K_M estimate at 8,192 context tokens, batch 1 and FP16 KV cache is 152.5 GiB including the safety margin. These are estimates, not measured guarantees.

Why does context length change memory use?

KV cache grows linearly with context length and batch size. We use KV heads for grouped-query attention and the recorded KV-cache layer count.

Does a green result guarantee compatibility?

No. Green means the estimated memory including safety headroom fits. Backend support, precision support, kernels and runtime allocation can differ. Confirm compatibility and measure your actual workflow.

Where do these estimates come from?

Architecture and file sizes are traced to model sources. The precision table identifies measured GGUF rates, block storage and nominal bit widths. The source data for Qwen3-235B-A22B-Thinking-2507 was last verified on 2026-11-20. Memory estimates use GiB.