PopularNewLLMMIT

DeepSeek R1

by DeepSeek

Jan 15, 2026•256K context•$0.25/M input•$1.00/M output

DeepSeek R1 has 52.8M downloads on HuggingFace. 671B MoE architecture with 37B active parameters. Features enhanced chain-of-thought reasoning. Scores 52.8% on SWE-bench Verified.

View on HuggingFace

Fleek Pricing

$0.0025/GPU-second

Context256K tokens

Estimated Token Cost

Input

$0.25/M

Output

$1.00/M

Based on 16,000 tokens/sec

vs CompetitorsSave 51%

Overview

Parameters

671B (MoE)

Architecture

Mixture of Experts

Context

256K

Provider

DeepSeek

Best For

Complex reasoningMathematical proofsCode generationScientific analysis

OpenAI Compatible

Drop-in replacement for OpenAI API. Just change the base URL.

Pay Per Second

Only pay for actual GPU compute time. No idle costs.

Enterprise Ready

99.9% uptime SLA, SOC 2 compliant, dedicated support.

Auto Scaling

Scales from zero to thousands of requests automatically.

Compare Pricing

	Fleek	Fireworks	Together	Baseten
Input	$0.25	$1.35	$3.00	—
Output	$1.00	$5.40	$7.00	—
Savings		70%	70%	—

Prices are per million tokens. Fleek pricing based on $0.0025/GPU-second.

Calculate Your Savings

See how much you'd save running DeepSeek R1 on Fleek

Model

DeepSeek R1

Compare To

Monthly Usage: 100M tokens

Quick Select

Your Fleek Cost

$100-167/mo

40.0K-66.7K GPU-sec × $0.0025

Fireworks AI

$338/mo

Your Savings61%

Annual Savings

$2.5K

See all models in full calculatororUpload your bill for custom analysis

Technical Specifications

Model Name	DeepSeek R1
Total Parameters	671B (MoE)
Active Parameters	37B
Architecture	Mixture of Experts
Context Length	256K tokens
Inference Speed	16,000 tokens/sec
Provider	DeepSeek
Release Date	Jan 15, 2026
License	MIT
HuggingFace	https://huggingface.co/deepseek-ai/DeepSeek-R1