DeepSeek
DeepSeek R1 API
DeepSeek R1 is an open reasoning model that punches at the level of much more expensive proprietary models. Like o3, it does extended chain-of-thought before answering — but at ~20× lower cost than o3.
- Input
- $0.55
- Output
- $2.19
- Context
- 128K
- Vision
- Text only
/ 1M tokens
/ 1M tokens
tokens
Top use cases
- Cost-sensitive math and code reasoning
- Open-source research workflows
- Self-hostable fallback for reasoning tasks
- Bulk evaluation and grading
Use DeepSeek R1 in 30 seconds
ModelServer is OpenAI-compatible. Point your existing OpenAI SDK at modelserver.dev/v1 and set the model name to deepseek-r1.
deepseek-r1.py
from openai import OpenAI
client = OpenAI(
api_key="sk-modelserver-...",
base_url="https://modelserver.dev/v1",
)
response = client.chat.completions.create(
model="deepseek-r1",
messages=[
{"role": "user", "content": "Hello, DeepSeek R1!"}
],
)
print(response.choices[0].message.content)Frequently asked questions
- How much does DeepSeek R1 cost?
- DeepSeek R1 is priced at $0.55 per 1M input tokens and $2.19 per 1M output tokens via ModelServer. ModelServer adds a flat 5.5% platform fee on top — no markups on individual tokens, no monthly minimum.
- What is the DeepSeek R1 context window?
- DeepSeek R1 supports a 128K token context window. You can put roughly 96,000 words in a single prompt.
- Is DeepSeek R1 OpenAI-compatible via ModelServer?
- Yes. Point your OpenAI SDK base_url to https://modelserver.dev/v1 and set model="deepseek-r1". Existing OpenAI-SDK code works without modification.
- Who is DeepSeek R1 best for?
- Reasoning workloads where price-per-token matters more than the brand.
- Does DeepSeek R1 support vision input?
- No. DeepSeek R1 is text-only. For multimodal use cases consider Claude Sonnet 4, GPT-4o, or Gemini 2.5 Pro.