vanchin/deepseek-r1

Updated at:

DeepSeek-R1 is a 671B-parameter Mixture of Experts (MoE) reasoning model open-sourced by DeepSeek in January 2025, activating only 37B parameters during inference. It is the first 100B+ parameter model trained purely via reinforcement learning without supervised fine-tuning, enabling the natural emergence of Chain-of-Thought (CoT). By incorporating cold-start data prior to RL, the model resolves the repetition and language mixing issues of R1-Zero, achieving performance on par with OpenAI o1 across math, coding, and reasoning tasks.

Inference Service Provider

The inference service provider for vanchin/deepseek-r1 is Kuaishou Vanchin.

Model Capabilities

CapabilitySupportCapabilitySupport

Input Modality

Text

Output Modality

Text

Model Experience

Unsupported

Function Calling

Supported

Structured Outputs

Unsupported

Web Search

Unsupported

Prefix Completion

Supported

Context Caching

Supported

Batch Inference

Unsupported

Fine-tuning

Unsupported

Context Limits

ParameterValueParameterValue

Max Input Length

98304

Max Output Length

32768

Context Window

131072

Max Input Length (Thinking Mode)

98304

Max Output Length (Thinking Mode)

32768

Pricing

This page only shows the original pricing for model API calls, excluding any limited-time promotions. Visit Model Studio Console for promotional offers.

China (Beijing)

Billing ItemPrice (CNY)Unit

Input

4

Per 1M tokens

Output

16

Per 1M tokens

Input(Implicit Cache)

1.6

Per 1M tokens

Rate Limits

China (Beijing)

ParameterValue

RPM (Requests Per Minute)

500

TPM (Tokens Per Minute)

1,000,000