DeepSeek
DeepSeek V3.2
Frontier
Jan 2026Veröffentlicht128K TokenKontextMITLizenz90 HerausragendQualität
DeepSeek V3.2 (671B parameters) requires approximately 411.6 GB of VRAM with Q4_K_M quantization. As a Mixture of Experts model with 37B active parameters, it uses less memory than its total parameter count suggests. For the best balance of quality and speed, we recommend hardware with at least 474 GB of VRAM.
Loslegen
— kopieren & einfügen, um lokal auszuführenCopy-paste commands to run DeepSeek V3.2 on your machine.
Run
ollama run deepseek-v3.2Quick specs
Parameters671B (37B active)
Architecturemoe (MoE)
Context128K tokens
Modalitytext
Min RAM261.7 GB
Rec. RAM409.3 GB (Q4_K_M)
LicenseMIT
FamilyDeepSeek
✓ Code✓ Chat✓ Reasoning
About this model
- •671B total / 37B active MoE
- •DeepSeek Sparse Attention
- •128K context
- •MIT license
- •67.8% SWE-bench Verified
Verwandte Modelle
Quantisierungsoptionen
VRAM-Schätzungen nach Quantisierungsstufe
No hardware detected — fit column shows raw VRAM estimates
| Quant | Bits | VRAM | Quality | Fit |
|---|---|---|---|---|
Q2_K | 2 | 261.7 GB | Low | — |
Q3_K_S | 3 | 328.8 GB | Low | — |
NVFP4 | 4 | 375.8 GB | Medium | — |
Q4_K_M | 4 | 409.3 GB | Medium | — |
Q5_K_M | 5 | 483.1 GB | High | — |
Q6_K | 6 | 550.2 GB | High | — |
Q8_0 | 8 | 718.0 GB | Very High | — |
F16 | 16 | 1375.6 GB | Maximum | — |
Quality benchmarks
DeepSeek V3.2 benchmark scores
Coding
SWE-bench Verified—
HumanEval+—
Aider Polyglot—
LiveCodeBench74.1%
Reasoning
MMLU-Pro85.0%
GPQA Diamond—
MATH-500—
ARC Challenge—
Source: official · 2025-12-01
Hardware-Kompatibilität
Eignungsschätzungen für alle Hardware
Computing compatibility...
Speicheraufschlüsselung
Reference: RTX 2060 6GB
Weights409.3 GB
KV Cache0.5 GB
Runtime1.2 GB
Headroom0.6 GB
Häufig gestellte Fragen
FAQ — DeepSeek V3.2
Siehe auch