opendunes

R1 Distill Llama 70B

by DeepSeekReleased 23 January 2025

Llama 3.3 70B taught by DeepSeek R1's outputs, bringing R1-style reasoning to maths and code with 70 on AIME 2024, 94.5 on MATH-500 and a 1633 CodeForces rating, near far larger models

TakesText
GivesText

Reads about 8,1 pages · writes up to 7,3 pages

Parameters you can send

Standard OpenAI fields, grouped by what they control. Anything not listed is accepted and ignored.

Shaping the reply

How long, how varied, where to stop.

max_tokenstemperaturetop_ptop_kstopseedfrequency_penaltypresence_penaltyrepetition_penalty

Reasoning

Let it think before answering — and see the thinking.

reasoninginclude_reasoning

Questions about R1 Distill Llama 70B

  • Is R1 Distill Llama 70B still on sale?

    No. DeepSeek has retired it. Other DeepSeek models are on sale.

  • What is R1 Distill Llama 70B best for?

    Llama 3.3 70B taught by DeepSeek R1's outputs, bringing R1-style reasoning to maths and code with 70 on AIME 2024, 94.5 on MATH-500 and a 1633 CodeForces rating, near far larger models.

  • Does R1 Distill Llama 70B work in Arabic, French or Darja?

    DeepSeek names no languages for R1 Distill Llama 70B. Arabic, French and Darja: not stated.

  • Can it read PDFs, images or a whole book?

    It reads text. Up to 8K tokens in one request, about 8 pages. It writes up to 6 000 words at once.

  • Is R1 Distill Llama 70B open source?

    Yes, its weights are public. Open weights on Hugging Face

  • Who makes R1 Distill Llama 70B, and when was it released?

    DeepSeek. Released 23 January 2025. It knows the world up to 31 July 2024.

Other DeepSeek models, priced in dinars

See all DeepSeek models
DeepSeek V4.1 Flash 1017DeepSeek · Oct 17, 2026

#10 for 3D scenes · This is the 17 October 2026 snapshot of V4.1 Flash, the image-and-text model DeepSeek serves under its current Flash name. It keeps about a thousand pages in mind, replies up to about 393 pages, thinks first, uses your connected tools and fills in the exact form you give it. Pick it for agents and coding work that include screenshots, and when you want the newest Flash. The earlier DeepSeek V4.1 Flash is the snapshot before this one; V4 Pro 0813 is the larger model for hard reasoning.

Thinks firstUses your tools1 000 pages at once
≈0,15 DAper message
View model
DeepSeek V4.1 FlashDeepSeek · Sep 10, 2026

DeepSeek V4.1 Flash is the first DeepSeek model built on a new design, and it reads images as well as text, keeping about a thousand pages in mind. It thinks before answering, uses your connected tools and fills in the exact form you give it. Use it as the fast, everyday model for coding agents that also need to look at screenshots or diagrams.

Thinks firstUses your tools1 000 pages at once
≈0,11 DAper message
View model
DeepSeek V4 Flash 0731DeepSeek · Jul 31, 2026

DeepSeek's generally available V4 Flash, re-post-trained for coding, reasoning and agent work on the same 284B mixture-of-experts with 13B active, with tools, structured output and switchable reasoning

#27 of 119 for codingThinks firstUses your tools
≈0,070 DAper message
View model
DeepSeek V4 Flash 0423DeepSeek · Apr 24, 2026

DeepSeek V4 Flash is the first V4 release of the Flash size, built for fast answers over very long text. It reads about a thousand pages at once, thinks before answering when asked, uses your connected tools and fills in the exact form or table you give it. Pick it to run an agent or summarise a whole codebase quickly. DeepSeek V4 Flash 0731 is the retrained revision for coding and agents; V4 Pro is the larger model.

Thinks firstUses your tools1 000 pages at once
≈0,061 DAper message
View model