ai.canshu.com
中文
VRAM Calculator Advisor & TCO API Pricing Compare Models Leaderboard GPUs

Search models, GPUs, and tools

↑↓ Navigate · Enter Open · ★ Watchlist prioritized ai.canshu.com
DeepSeek
Open-Weight DeepSeek · MoE 671B / A37B · Text · FP8 Mixed

DeepSeek-R1 (671B-A37B MoE) Architecture Parameters, Benchmarks & VRAM

Full 671B MoE reasoning model with MLA KV-cache compression and 37B active parameters per token.

🧬 DeepSeek-R1 (671B-A37B MoE) Architecture & Tensor Specifications

Verified directly against HuggingFace (deepseek-ai/DeepSeek-R1) config.json & Model Card

HF Class: DeepseekV3ForCausalLM

1. Backbone & Dimensions

deepseek_v3
Total / Active Params
671B / 37B
Hidden Size (d_model)
7,168
FFN Intermediate Size
18,432
MoE Expert Dim
2,048
MoE Experts
256 Routed · Top-8 + 1 Shared
Activation / Norm
SiLU (SwiGLU) · ε=1e-6
Native Tensor Dtype
FP8 Mixed

2. Attention & Layer Layout

MLA
Num Hidden Layers
61 Layers
Hidden Layer Layout
61-Layer DeepSeekMoE + MLA (First 3 Dense FFN + 58 MoE Layers)
Attention Heads (Q / KV)
128 Q / 128 KV
Head Dimension
56
Linear / KV Compression
MLA Low-Rank KV Compression: kv_lora_rank=512 + qk_rope_head_dim=64
Vision Encoder / MTP
MTP 1L

3. Context, RoPE & Tokenizer

Tokenizer
Native Context Length
163,840 tokens
RoPE Configuration
YaRN Decoupled RoPE (θ=10,000 · qk_rope_dim=64 · 160K Context)
Vocab Size
129,280 (Untied)
Special Token IDs
BOS: 0 · EOS: 1

📋 DeepSeek-R1 (671B-A37B MoE) Official API & System Card Specifications

Verified against official DeepSeek developer documentation & pricing

API Model ID: deepseek-r1

1. Identity & Capabilities

API Spec
API Model ID
deepseek-r1
Release Date
2025-05-28
Weight License
Open-Weight (API Spec)
Native Reasoning
✅ Supported
Function Calling & JSON Schema
✅ 原生支持
Batch API 异步批处理
✅ Supported (50% off)

2. Context, Output & Modalities

Limits
Max Context Window
128,000 tokens (128K)
Max Output Tokens
16,000 tokens
Input Modalities
Text
Output Modalities
Text
Direct China Access
✅ Direct CN + RMB
Deployment Mode
Open-Weight + Cloud API

3. Official Pricing & Cache

CNY / 1M
Input / 1M Tokens
¥ 4
Cached Input / 1M
¥ 0.8 -80%
Output / 1M Tokens
¥ 16
Tiered Pricing Notes
Alibaba Bailian: Hosted on Alibaba Cloud Bailian (CN Direct)

🏆 Official Benchmark Results

Sourced from official System Card / Model Card & technical reports

Arena Elo ~1476 AA Index 48
📚 Knowledge
MMLU 90.8 %

Massive Multitask Language Understanding benchmark testing knowledge across 57 diverse subjects including STEM, humanities, social sciences, and professional domains.

📚 Knowledge
MMLU-Redux 92.9 %

An improved version of the MMLU benchmark featuring manually re-annotated questions to identify and correct errors in the original dataset. Provides more reliable evaluation metrics for language models by addressing dataset quality issues found in the original MMLU.

📚 Knowledge
MMLU-Pro 84 %

A more robust and challenging multi-task language understanding benchmark that extends MMLU by expanding multiple-choice options from 4 to 10, eliminating trivial questions, and focusing on reasoning-intensive tasks.

🧠 Reasoning
GPQA Diamond 71.5 %

PhD-level Google-proof science reasoning questions

🧠 Reasoning
MATH-500 97.3 %

MATH-500 is a subset of the MATH dataset containing 500 challenging competition mathematics problems from AMC 10, AMC 12, AIME, and other mathematics competitions.

🧠 Reasoning
AIME 2024 79.8 %

American Invitational Mathematics Examination 2024, consisting of 30 challenging mathematical reasoning problems from AIME I and AIME II competitions.

💻 Coding
LiveCodeBench 65.9 %

LiveCodeBench is a holistic and contamination-free evaluation benchmark for large language models for code.

🤖 Agentic
AlpacaEval 2.0 87.6 %

Length-controlled instruction-following win-rate benchmark

🎛️ Official Sampling & Reasoning Parameters

Recommended generation settings & API parameter constraints

Reasoning Recommended Config
temperature=0.6 top_p=0.95 max_tokens=32,768

💡 Keep temperature in 0.5–0.7 (0.6 best); avoid system prompts and place all instructions in user message.

⚡ DeepSeek-R1 (671B-A37B MoE) Interactive VRAM & Token Speed Calculator

Concurrency (Batch):
MoE 671B / A37B·61L (d=7168)·128Q/128KV·Ctx 128K
Runnable Models →
GPUs
96 GB VRAM·4000 GB/s·Safe Usable 94.1 GB
Out of VRAM (OOM)416.53 / 94.1 GBOOM
3. Weight Quantization
4. Per-Request Context16K (16,384 tok)
Verified: 2026-11-20·Roofline bandwidth/compute estimate (Batch=1)
Out of VRAM (OOM)
416.53/ 94.1 GB usable

Physical 96 GB · Load 442.7%

Theoretical Decode Speed
—tokens/s

Exceeds usable VRAM by 322.45 GB

Max Context at Batch=1
0K/ 128K

MLA compresses multi-batch KV footprint

Weights: 406.79 GBKV Cache (1×16K): 1.15 GBOverhead: 8.59 GB
⛔ Cannot Load
Reference Launch CLI:
OLLAMA_CONTEXT_LENGTH=16384 ollama run deepseek-r1:671b

Insufficient Local VRAM? Ways to Run DeepSeek-R1 (671B-A37B MoE)

Rent an hourly high-VRAM cloud GPU pod, or check the 12-month Buy vs Cloud vs API break-even curve.

📊 DeepSeek-R1 (671B-A37B MoE) Quantization VRAM Breakdown (16K Context)

Quantization Bits/Weight Weights KV Cache (16K) Total VRAM Quality Note
FP16 / BF16 16 bpw 1342 GB 1.15 GB 1370.44 GB Lossless Full Precision
FP8 / Q8_0 8.5 bpw 712.94 GB 1.15 GB 728.8 GB 99.9% Near-Lossless
GGUF Q6_K 6.56 bpw 550.22 GB 1.15 GB 562.83 GB Ultra High Quality
GGUF Q5_K_M 5.69 bpw 477.25 GB 1.15 GB 488.4 GB High Balanced Quality
Q4_K_M / AWQ 4.85 bpw 406.79 GB 1.15 GB 416.53 GB Sweet Spot (Recommended)
GGUF IQ4_XS 4.25 bpw 356.47 GB 1.15 GB 365.2 GB Compact 4-bit
GGUF Q3_K_M 3.91 bpw 327.95 GB 1.15 GB 336.11 GB Moderate Perplexity Loss
GGUF Q2_K 3.15 bpw 264.21 GB 1.15 GB 271.09 GB Extreme Compression

🖥️ GPU & Apple Silicon Compatibility for DeepSeek-R1 (671B-A37B MoE) (Q4_K_M)

Frequently Asked Questions

What are the exact architecture parameters of DeepSeek-R1 (671B-A37B MoE)?

DeepSeek-R1 (671B-A37B MoE) uses DeepseekV3ForCausalLM with 61 layers (61-Layer DeepSeekMoE + MLA (First 3 Dense FFN + 58 MoE Layers)), hidden_size=7168, 128 Q heads / 128 KV heads (head_dim=56), and vocab_size=129,280.

How much VRAM does DeepSeek-R1 (671B-A37B MoE) require?

At 4-bit quantization (Q4_K_M, 16K context), DeepSeek-R1 (671B-A37B MoE) requires 416.53 GB VRAM. Full FP16 requires 1370.44 GB.