Alibaba Cloud · qwen/qwq-32b
| Provider | Alibaba Cloud |
|---|---|
| Family | qwq |
| Type | llm-reasoning |
| Status | active |
| Released | 2025-03-05 |
| Updated | 2025-03-11 |
| Parameters | 32.8B |
| Open weights | yes |
| Relationship | Model |
|---|---|
| Is a finetune of | Qwen/Qwen2.5-32B |
Ordered by predicted speed. Bandwidth sets decode rate; memory decides whether it runs at all.
| Device | Memory | Bandwidth | Best quality | Weights | tok/s | Fastest |
|---|---|---|---|---|---|---|
| Cerebras WSE-3 | 44.0 GB | 21000000.0 GB/s | fp8 | 32.76 GB | ~448664.9 | ~897329.7 q4 |
| NVIDIA GB200 Grace Blackwell Superchip | 372.0 GB | 16000.0 GB/s | bf16 | 65.53 GB | ~170.9 | ~683.7 q4 |
| AMD Instinct MI355X | 288.0 GB | 8000.0 GB/s | bf16 | 65.53 GB | ~85.5 | ~341.8 q4 |
| NVIDIA B200 | 180.0 GB | 7700.0 GB/s | bf16 | 65.53 GB | ~82.3 | ~329.0 q4 |
| Google TPU7x (Ironwood) | 192.0 GB | 7380.0 GB/s | bf16 | 65.53 GB | ~78.8 | ~315.3 q4 |
| AMD Instinct MI325X | 256.0 GB | 6000.0 GB/s | bf16 | 65.53 GB | ~64.1 | ~256.4 q4 |
| AMD Instinct MI300X | 192.0 GB | 5300.0 GB/s | bf16 | 65.53 GB | ~56.6 | ~226.5 q4 |
| NVIDIA H200 SXM | 141.0 GB | 4800.0 GB/s | bf16 | 65.53 GB | ~51.3 | ~205.1 q4 |
| NVIDIA H100 NVL | 94.0 GB | 3900.0 GB/s | bf16 | 65.53 GB | ~41.7 | ~166.6 q4 |
| NVIDIA H100 SXM | 80.0 GB | 3350.0 GB/s | fp8 | 32.76 GB | ~71.6 | ~143.1 q4 |
| AMD Instinct MI250X | 128.0 GB | 3200.0 GB/s | bf16 | 65.53 GB | ~34.2 | ~136.7 q4 |
| Google TPU v5p | 95.0 GB | 2765.0 GB/s | bf16 | 65.53 GB | ~29.5 | ~118.1 q4 |
| NVIDIA A100 80GB SXM | 80.0 GB | 2039.0 GB/s | fp8 | 32.76 GB | ~43.6 | ~87.1 q4 |
| NVIDIA H100 PCIe | 80.0 GB | 2000.0 GB/s | fp8 | 32.76 GB | ~42.7 | ~85.5 q4 |
| NVIDIA GeForce RTX 5090 | 32.0 GB | 1792.0 GB/s | q5 | 20.48 GB | ~61.3 | ~76.6 q4 |
| Google TPU v6e (Trillium) | 32.0 GB | 1638.0 GB/s | q5 | 20.48 GB | ~56.0 | ~70.0 q4 |
| AMD Instinct MI210 | 64.0 GB | 1600.0 GB/s | fp8 | 32.76 GB | ~34.2 | ~68.4 q4 |
| NVIDIA A100 40GB SXM | 40.0 GB | 1555.0 GB/s | q6 | 24.57 GB | ~44.3 | ~66.4 q4 |
| Google TPU v4 | 32.0 GB | 1200.0 GB/s | q5 | 20.48 GB | ~41.0 | ~51.3 q4 |
| NVIDIA GeForce RTX 4090 | 24.0 GB | 1008.0 GB/s | q4 | 16.38 GB | ~43.1 | ~43.1 q4 |
| AMD Radeon RX 7900 XTX | 24.0 GB | 960.0 GB/s | q4 | 16.38 GB | ~41.0 | ~41.0 q4 |
| NVIDIA GeForce RTX 3090 | 24.0 GB | 936.0 GB/s | q4 | 16.38 GB | ~40.0 | ~40.0 q4 |
| NVIDIA L40S | 48.0 GB | 864.0 GB/s | fp8 | 32.76 GB | ~18.5 | ~36.9 q4 |
| Apple M3 Ultra | 512.0 GB | 819.0 GB/s | bf16 | 65.53 GB | ~8.7 | ~35.0 q4 |
| Apple M2 Ultra | 192.0 GB | 800.0 GB/s | bf16 | 65.53 GB | ~8.5 | ~34.2 q4 |
| Apple M4 Max | 128.0 GB | 546.0 GB/s | bf16 | 65.53 GB | ~5.8 | ~23.3 q4 |
| Apple M1 Max | 64.0 GB | 400.0 GB/s | fp8 | 32.76 GB | ~8.5 | ~17.1 q4 |
| Apple M2 Max | 96.0 GB | 400.0 GB/s | bf16 | 65.53 GB | ~4.3 | ~17.1 q4 |
| Apple M3 Max | 128.0 GB | 400.0 GB/s | bf16 | 65.53 GB | ~4.3 | ~17.1 q4 |
| NVIDIA L4 | 24.0 GB | 300.0 GB/s | q4 | 16.38 GB | ~12.8 | ~12.8 q4 |
| Apple M4 Pro | 64.0 GB | 273.0 GB/s | fp8 | 32.76 GB | ~5.8 | ~11.7 q4 |
| Apple M4 | 32.0 GB | 120.0 GB/s | q5 | 20.48 GB | ~4.1 | ~5.1 q4 |
Taken from the card's availability section.
| Platform | Id on that platform |
|---|---|
| Hugging Face | Qwen/QwQ-32B |
| Benchmark | Evaluated as | Score | Evidence date | Source kind | |
|---|---|---|---|---|---|
| aa_lcr | QwQ-32B | 26.67% | 2026-09-10 evaluated | independent evaluator | source |
| gpqa_diamond | QwQ-32B | 59.29% | 2026-09-10 evaluated | independent evaluator | source |
This card reports no benchmark scores yet.
| Benchmark | Catalogue standing | Score |
|---|