Models / DeepSeek

Model vendor

DeepSeek

DeepSeek publishes open-weight frontier models: general-purpose V-series models and R-series reasoning models, both at very low prices.

15Models routed
+37%Average vs list price
2Channels serving DeepSeek

DeepSeek models on MirAPI

ModelTypeContextInputOutputCache r / wOff listChannels
DeepSeek V4.1 Flash 1017newText1.049M$0.15$0.60— / —soon—
DeepSeek V4 Flash Vision ExpnewText1.049M$0.22$0.65— / —soon—
DeepSeek V4 Flash 0423Text1.049M$0.09$0.18— / —n/a—
DeepSeek V4 Pro 0423Text1.049M$0.21$0.42— / —n/a—
DeepSeek V3.2Text164K$0.28$0.42— / —n/a—
DeepSeek V3.1 TerminusText164K$0.27$1.00— / —n/a—
DeepSeek V3.2 ExpText164K$0.27$0.41— / —n/a—
DeepSeek V3.1Text164K$0.25$0.95— / —n/a—
R1 0528Text164K$0.50$2.15— / —n/a—
DeepSeek V3 0324Text164K$0.29$1.14— / —n/a—
R1Text64K$0.70$2.50— / —n/a—
DeepSeek V3Text164K$0.32$0.89— / —n/a—
Browse the full catalog
Migration

A drop-in replacement for the OpenAI API

MirAPI accepts the OpenAI wire format, so the official SDKs, LangChain, agent frameworks and your own wrapper all work as they are. Change the base URL and use model IDs as they appear in our catalog:

Endpoints
/v1/chat/completionsChat & tool calls
import os
from openai import OpenAI

client = OpenAI(
    base_url="https://api.mirapi.ai/v1",   # changed
    api_key=os.environ["MIRAPI_API_KEY"],   # changed
)

resp = client.chat.completions.create(
    model="deepseek-v4-pro-0813",
    messages=[{"role": "user", "content": "Hello"}],
)

One key covers DeepSeek and 10 more vendors.

1M free tokens to start. No card required.