Pass IndexThe State of AISign in

DeepSeek-R1-Distill-Llama-8B

DeepSeek-R1-Distill-Llama-8B is an 8B model based on Llama-3.1-8B and fine-tuned on reasoning traces generated by DeepSeek-R1, demonstrating that reasoning patterns of larger models can be distilled into smaller ones.

text → text · made by DeepSeek

$0.025$0.025per Mtok in / out
Nscale · 3 sellers

Sold by 3 ways

SellerLaneRate
Nscale aggregatorstandard$0.025per Mtok in$0.025per Mtok outraw.githubusercontent.com · read 2026-09-16
Hugging Face Inference Providers aggregatornscale$0.05per Mtok in$0.05per Mtok outrouter.huggingface.co · read 2026-08-25
Fireworks AI aggregatorstandard$0.2per Mtok in$0.2per Mtok outraw.githubusercontent.com · read 2026-09-16

About

DeepSeek-R1-Distill-Llama-8B — a text model from DeepSeek, sold by 3 companies from $0.025 in and $0.025 out per million tokens.

It takes text and returns text, with a context window of 128,000 tokens. It was published in 2025. Its sellers say it can reason step by step and call a tool. The catalogue files it under evaluate. Three companies sell it. The cheapest is $0.025 in and $0.025 out per million tokens at Nscale. Beside the standard rate there is a separately routed lane.

Every current figure

Maker
DeepSeek
Register
model
Takes
text
Returns
text
Context
128,000 tokens
Longest answer
16,384 tokens
Published
2025
Parameters
8 billion · read from its own name
Licence
not read
Sellers
3
Maker's own price
not read
Price
$0.025 in and $0.025 out per million tokens — Nscale

Known as 1 name

deepseek-ai/DeepSeek-R1-Distill-Llama-8B