← Back to models

DeepSeek-V3.1

DeepSeek · Mixture-of-Experts (MoE) · License: MIT

DeepSeek's largest general-purpose model. Later DeepSeek releases (V3.1, R1) unified their licensing under plain MIT, unlike the original December 2024 V3 release which split code (MIT) and weights (a separate custom license).

Memory requirement — the visual calculator

In real deployments, memory needs also depend on numeric precision (e.g. FP16 vs INT8 vs INT4), which changes bytes-per-parameter. This store uses a simplified, transparent formula so beginners can follow the logic: (parameters in billions × 1 GB) + 20% overhead for the runtime, active requests, and context memory.

Model weights
671 GB
+ 20% overhead
134.2 GB
= Total required
805.2 GB

Calculation: 671B parameters × 1 GB = 671 GB, plus 20% (134.2 GB) = 805.2 GB total The dedicated memory built into the graphics card. An AI model's entire set of "weights" (its learned parameters) must fit into this memory before the card can run it. If a model needs more VRAM than one card has, you need multiple cards working together, or a card with more memory. needed. Note: even though only 37B parameters ( A model design where only a fraction of the model's parameters ("active parameters") are used to process any single piece of text, even though the full model is much larger. This makes the model faster to run than its total size suggests, but you still need enough memory to hold the entire model, not just the active part. ) are active per token, the full 671B must still be loaded into memory.

Cheapest way to run this model today
1× NVIDIA DGX B200 (1440 GB total VRAM, ~$387,500). See the full recommendation →

Leaderboard

Added to the LMArena text leaderboard Aug 29, 2025, debuting at #8 overall.

Sources