AI Potluck
Model components / Fine-tuned / chat models

DeepSeek-R1

DeepSeek

Previous-generation reasoning-first model (January 2025) using large-scale GRPO RL to develop chain-of-thought before answering. 671B MoE plus 1.5B–70B distilled variants. Superseded by DeepSeek-R2 (April 2026), which inverts the architecture thesis; R2 is a 32B dense MIT model that runs on a single 24GB consumer GPU and reaches 92.7% AIME 2025. R1 is still by some distance the most-downloaded open reasoning model on the Hub.

DeepSeek-R1 (671B total / 37B active MoE, 128K context), reasoning model RL+SFT post-trained on DeepSeek-V3-Base (cold-start data + RL + SFT pipeline; R1-Zero is pure-RL variant); arXiv 2501.12948, released 2025-01-22. The earlier record put monthly downloads at 4.3M; the card now reads 8.3M, so the figure was dropped from the description rather than re-pinned. Verified 2026-08-13 via the HF model card.

Openness

3 high confidence
3.0
weights
open(MIT, on HF
data
closed(RL/SFT training data not released)
code
partial(inference
license
MIT(OSI-style permissive, no use restrictions)

MIT-licensed open weights that explicitly permit distillation, but the RL and SFT post-training data and the full training code are not released, so this is open weights rather than open source.

  • https://huggingface.co/deepseek-ai/DeepSeek-R1 recorded 2026-08-13

    `license:mit` in the repo tags and `"gated":false` in the embedded repo state, so the 671B-A37B MoE weights download without a barrier; the card carries evaluation tables and local-inference instructions only, with no RL training pipeline and no post-training mixture released

Adoption

4 high confidence
4.0

The base R1 repository reads 8,316,405 Hugging Face downloads in the trailing 30 days, and with the declared R1-0528 SKU (148,234) the tier sums to 8,464,639 - inside the 1M-10M band and near the top of it. That count is before the very large family of R1 distills and the DeepSeek app and API surface, none of which it captures. The R1 release was a landmark open-reasoning event and the download volume still reflects it.

Capability

4 high confidence
4.0

Frontier open reasoning model at its Jan 2025 release (o1-class: matched/beat o1 on AIME, MATH-500, LiveCodeBench). As of June 2026 it is surpassed by DeepSeek-V3.2-Speciale/V4 and other 2026 frontier reasoners, so a 4 rather than a category-5 in mid-2026 terms. Figures are DeepSeek-reported on the HF card.

Verified 2026-08-13