NAVER · naver/hyperclovax-seed-think-32b
| Provider | NAVER |
|---|---|
| Family | hyperclova |
| Type | llm-reasoning |
| Status | active |
| Released | 2025-12-23 |
| Updated | 2026-07-14 |
| Parameters | 33.3B |
| Open weights | yes |
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 | q6 | 24.99 GB | ~588351.7 | ~882527.5 q4 |
| NVIDIA GB200 Grace Blackwell Superchip | 372.0 GB | 16000.0 GB/s | bf16 | 66.63 GB | ~168.1 | ~672.4 q4 |
| AMD Instinct MI355X | 288.0 GB | 8000.0 GB/s | bf16 | 66.63 GB | ~84.1 | ~336.2 q4 |
| NVIDIA B200 | 180.0 GB | 7700.0 GB/s | bf16 | 66.63 GB | ~80.9 | ~323.6 q4 |
| Google TPU7x (Ironwood) | 192.0 GB | 7380.0 GB/s | bf16 | 66.63 GB | ~77.5 | ~310.1 q4 |
| AMD Instinct MI325X | 256.0 GB | 6000.0 GB/s | bf16 | 66.63 GB | ~63.0 | ~252.2 q4 |
| AMD Instinct MI300X | 192.0 GB | 5300.0 GB/s | bf16 | 66.63 GB | ~55.7 | ~222.7 q4 |
| NVIDIA H200 SXM | 141.0 GB | 4800.0 GB/s | bf16 | 66.63 GB | ~50.4 | ~201.7 q4 |
| NVIDIA H100 NVL | 94.0 GB | 3900.0 GB/s | bf16 | 66.63 GB | ~41.0 | ~163.9 q4 |
| NVIDIA H100 SXM | 80.0 GB | 3350.0 GB/s | fp8 | 33.31 GB | ~70.4 | ~140.8 q4 |
| AMD Instinct MI250X | 128.0 GB | 3200.0 GB/s | bf16 | 66.63 GB | ~33.6 | ~134.5 q4 |
| Google TPU v5p | 95.0 GB | 2765.0 GB/s | bf16 | 66.63 GB | ~29.0 | ~116.2 q4 |
| NVIDIA A100 80GB SXM | 80.0 GB | 2039.0 GB/s | fp8 | 33.31 GB | ~42.8 | ~85.7 q4 |
| NVIDIA H100 PCIe | 80.0 GB | 2000.0 GB/s | fp8 | 33.31 GB | ~42.0 | ~84.1 q4 |
| NVIDIA GeForce RTX 5090 | 32.0 GB | 1792.0 GB/s | q5 | 20.82 GB | ~60.2 | ~75.3 q4 |
| Google TPU v6e (Trillium) | 32.0 GB | 1638.0 GB/s | q5 | 20.82 GB | ~55.1 | ~68.8 q4 |
| AMD Instinct MI210 | 64.0 GB | 1600.0 GB/s | fp8 | 33.31 GB | ~33.6 | ~67.2 q4 |
| NVIDIA A100 40GB SXM | 40.0 GB | 1555.0 GB/s | q6 | 24.99 GB | ~43.6 | ~65.3 q4 |
| Google TPU v4 | 32.0 GB | 1200.0 GB/s | q5 | 20.82 GB | ~40.3 | ~50.4 q4 |
| NVIDIA GeForce RTX 4090 | 24.0 GB | 1008.0 GB/s | q4 | 16.66 GB | ~42.4 | ~42.4 q4 |
| AMD Radeon RX 7900 XTX | 24.0 GB | 960.0 GB/s | q4 | 16.66 GB | ~40.3 | ~40.3 q4 |
| NVIDIA GeForce RTX 3090 | 24.0 GB | 936.0 GB/s | q4 | 16.66 GB | ~39.3 | ~39.3 q4 |
| NVIDIA L40S | 48.0 GB | 864.0 GB/s | fp8 | 33.31 GB | ~18.2 | ~36.3 q4 |
| Apple M3 Ultra | 512.0 GB | 819.0 GB/s | bf16 | 66.63 GB | ~8.6 | ~34.4 q4 |
| Apple M2 Ultra | 192.0 GB | 800.0 GB/s | bf16 | 66.63 GB | ~8.4 | ~33.6 q4 |
| Apple M4 Max | 128.0 GB | 546.0 GB/s | bf16 | 66.63 GB | ~5.7 | ~22.9 q4 |
| Apple M1 Max | 64.0 GB | 400.0 GB/s | fp8 | 33.31 GB | ~8.4 | ~16.8 q4 |
| Apple M2 Max | 96.0 GB | 400.0 GB/s | bf16 | 66.63 GB | ~4.2 | ~16.8 q4 |
| Apple M3 Max | 128.0 GB | 400.0 GB/s | bf16 | 66.63 GB | ~4.2 | ~16.8 q4 |
| NVIDIA L4 | 24.0 GB | 300.0 GB/s | q4 | 16.66 GB | ~12.6 | ~12.6 q4 |
| Apple M4 Pro | 64.0 GB | 273.0 GB/s | fp8 | 33.31 GB | ~5.7 | ~11.5 q4 |
| Apple M4 | 32.0 GB | 120.0 GB/s | q5 | 20.82 GB | ~4.0 | ~5.0 q4 |
Taken from the card's availability section.
| Platform | Id on that platform |
|---|---|
| Hugging Face | naver-hyperclovax/HyperCLOVAX-SEED-Think-32B |
| Benchmark | Evaluated as | Score | Evidence date | Source kind | |
|---|---|---|---|---|---|
| aa_lcr | HyperCLOVA X SEED Think (32B) | 13.67% | 2026-09-10 evaluated | independent evaluator | source |
| critpt | HyperCLOVA X SEED Think (32B) | 0.0% | 2026-09-10 evaluated | independent evaluator | source |
| gpqa_diamond | HyperCLOVA X SEED Think (32B) | 61.52% | 2026-09-10 evaluated | independent evaluator | source |
This card reports no benchmark scores yet.
| Benchmark | Catalogue standing | Score |
|---|