DE
DeepSeek
Language model · China ·Jan 2025

DeepSeek-R1

Language Open (Restricted)
671B
Parameters
27
Benchmarks

About DeepSeek-R1

DeepSeek-R1 is an AI model developed by DeepSeek, in the language category, released in 2025, made available as an open-weights model with 671B params.

On this page you'll find DeepSeek-R1's full specifications, including 27 benchmark results. Review provider pricing and benchmark scores below, or compare DeepSeek-R1 head-to-head with other language models.

Links

Run DeepSeek-R1 locally — what it takes

DeepSeek-R1 (671B) is too large to run on a single consumer or Apple machine. At Q4 its weights alone are about 376 GB — it needs a multi-GPU / datacenter setup:

  • ≈ 2× AMD Instinct MI325X
  • ≈ 3× NVIDIA B200

Lower quantization (Q3/Q2) or CPU + system-RAM offload (e.g. ktransformers) reduces the requirement at the cost of speed.

Software support

✓ measured · · compatible · — not supported. Informational only — speed is hardware-based.

Community

Would you run DeepSeek-R1 again?
No votes yet — be the first.
Did it run? — community reports by hardware

No reports yet. Own this model on some hardware? Be the first to confirm it runs.

Discussion

No comments yet. Start the discussion.

DeepSeek-R1 benchmark scores
Benchmark Category Score Variant
Aider Polyglot code 53.3% cited-magistral
Codeforces Elo code 2,029 rating-cited-phi4r
LiveCodeBench v5 code 65.9% cited-magistral
LiveCodeBench v5 code 65.9% lcb-aug24-jan25-cited-phi4r
LiveCodeBench v5 code 61.01%
AIME 2024 math 86.7% cons@64-cited-exaone
AIME 2024 math 79.8% cited-magistral
AIME 2024 math 78.7% cited-phi4r
AIME 2024 math 74.38%
AIME 2025 math 80% cons@64-cited-exaone
AIME 2025 math 70.4% cited-phi4r
AIME 2025 math 70% cited-magistral
AIME 2025 math 66.8% pass@1-cited-exaone
AIME 2025 math 65.21%
AIME 2025 math 65% cited-s1
HMMT February 2025 math 47.08%
HMMT February 2025 math 41.7% cited-phi4r
MATH-500 math 97.3% cited-s1
MATH-500 math 97.3% cited-magistral
MATH-500 math 97.3% pass@1-cited-exaone
Omni-MATH math 85 cited-phi4r
Omni-MATH-HARD math 51.33%
GPQA Diamond reasoning 71.5% cited-s1
GPQA Diamond reasoning 71.5% cited-magistral
GPQA Diamond reasoning 71.08%
Humanity's Last Exam reasoning 8.6% text-only-cited-magistral
Humanity's Last Exam reasoning 8.5%

More models like DeepSeek-R1

Frequently asked questions

What is DeepSeek-R1?

DeepSeek-R1 is an AI model developed by DeepSeek, in the language category, released in 2025. It is tracked on GenAIList with its specifications, benchmark scores and provider pricing.

Who created DeepSeek-R1?

DeepSeek-R1 was developed by DeepSeek and released in 2025.

Is DeepSeek-R1 open source or proprietary?

DeepSeek-R1 ships as an open-weights model: you can download and self-host the weights, though the licence may place some restrictions on use.

How much does DeepSeek-R1 cost?

Pricing for DeepSeek-R1 depends on the provider. See the providers table on this page for the latest API rates.

How does DeepSeek-R1 perform on benchmarks?

DeepSeek-R1 is benchmarked across 27 evaluations on GenAIList, including Aider Polyglot (53.3%). See the full benchmark table below and compare it with other models.

No videos yet. Know a good one? Add it below.

Know a great video about DeepSeek-R1?