DeepSeek R1

deepseek-ai · DeepSeek-R1 family · released 2025-01-20 · mit license

Open frontier reasoning model. Massive (671B/37B-active); realistically multi-GPU or heavy offload, or a low-bit GGUF.

Key specs

TypeLocal open-weight
Parameters684.53B total · MoE, 37B active
Architecturedeepseek_v3
Context window164K tokens
Knowledge cutoff2024-10-01
Modalitiestext
Recommended backends
Minimum viable rigMulti-GPU / heavy offload (FP8 weights ~700GB class)

Benchmark scores

GPQA Diamond71.5%
SWE-bench Verified49.2%
AIME79.8%
MMLU-Pro84%
BFCL v3 (tool use)
Composite score6.9
Community rating5.0★ (1 reviews, 0 net votes)

VRAM & disk per quantization

QuantVRAMDiskRAMContext
Q4_K_M398.5 GB397 GB448 GB164K

API pricing (per 1M tokens)

ProviderInputOutputFree tier
OpenRouter$0.5$2.18Yes
DeepSeek$0.55$2.19No
Together AI$3$7No

Strengths & weaknesses

Strengths: SOTA open reasoning (AIME 2024 79.8, MATH-500 97.3); Only 37B of 671B active per token (+ MLA KV compression) — cheap to serve for its class; Permissive MIT license incl. distillation

Weaknesses: Prompt-sensitive; can emit empty think blocks; Weaker factuality (SimpleQA 30.1); Very large — needs multi-GPU or heavy offload