Lunomi
Sign in Get an API key

DeepSeek V4.1 Flash1M-token contextOpenAI & Anthropic SDKs

The private, prepaid DeepSeek API

DeepSeek V4.1 Flash with a million-token context, for the OpenAI and Anthropic SDKs, from $0.22 per million input tokens at launch pricing. We don't log your prompts or replies.

Get an API key Read the docs

Prepaid from $5 in Bitcoin. No subscription.

Works with your tools

Anything that lets you set a base URL and an API key. The highlighted lines are all that change.

curl https://api.lunomi.io/v1/chat/completions \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model": "deepseek-v4.1-flash", "messages": [{"role": "user", "content": "Hello"}]}'

Plain HTTP works too: this is the OpenAI format; the Anthropic one is POST https://api.lunomi.io/v1/messages.

from openai import OpenAI

client = OpenAI(
    base_url="https://api.lunomi.io/v1",
    api_key="YOUR_API_KEY",
)

Streaming, tool calls and JSON mode work as they do with OpenAI.

import anthropic

client = anthropic.Anthropic(
    base_url="https://api.lunomi.io",
    api_key="YOUR_API_KEY",
)

The Anthropic format calls /v1/messages, so its base URL has no /v1.

export ANTHROPIC_BASE_URL="https://api.lunomi.io"
export ANTHROPIC_AUTH_TOKEN="YOUR_API_KEY"
export ANTHROPIC_MODEL="deepseek-v4.1-flash"

# Agents on the OpenAI API read these
export OPENAI_BASE_URL="https://api.lunomi.io/v1"
export OPENAI_API_KEY="YOUR_API_KEY"

Coding agents built on either API pick these up from the environment.

Provider:  OpenAI-compatible
Base URL:  https://api.lunomi.io/v1
API key:   YOUR_API_KEY
Models:    deepseek-v4.1-flash, deepseek-v4.1-flash-nothink

Cursor, Open WebUI, LibreChat and similar apps take a custom OpenAI-compatible provider. Most list the models from /v1/models.

models:
  - name: DeepSeek V4.1 Flash
    provider: openai
    model: deepseek-v4.1-flash
    apiBase: https://api.lunomi.io/v1
    apiKey: YOUR_API_KEY

Continue treats it as an OpenAI provider with your own apiBase.

Context
1,048,576
tokens per request
Max output
32,768
tokens per answer
Input
$0.22
per 1M tokens
Cached input
$0.045
per 1M tokens
Output
$0.75
per 1M tokens
Prompts logged
No
token counts and cost only

Launch pricing: it may change after our benchmarks, with notice. Full pricing, a cost estimator and the rate limits

One model, DeepSeek V4.1 Flash (release 1017), under two names: with its reasoning step, or without it for faster, shorter answers.

deepseek-v4.1-flash
On: thinks before answering
deepseek-v4.1-flash-nothink
Off: answers directly
Speaks
  • Chat completions (OpenAI)
  • Messages (Anthropic)
  • Streaming
  • Tool calls
  • JSON mode
  • Automatic prompt caching
Works with
  • OpenAI SDK
  • Anthropic SDK
  • Cursor
  • Continue
  • Open WebUI
  • LibreChat
  • coding agents

How a request travels

  1. 1Your app sends a request

    Through the OpenAI or Anthropic SDK, with your key.

  2. 2It crosses the internet, encrypted

    HTTPS all the way to our gateway.

  3. 3The gateway checks it

    Key and credit checked, then it goes to a GPU over our private network.

  4. 4Prefill reads your prompt

    The whole prompt, in one parallel pass. This is what input pays for.

  5. 5The KV cache keeps that work

    In GPU memory, for your account only. Repeats cost 80% less.

  6. 6Decode writes the answer

    One token at a time. This is what output pays for.

  7. 7Back to you

    Tokens stream back the same way. We log counts and cost, not text.

A request goes from your app across the internet to our gateway, over a private network to a GPU server, and the answer comes back the same way. Internet · HTTPS Private network Your app Gateway key ✓ credit ✓ → GPU GPU servers
Illustration. Not a specific GPU model.

From sign-up to your first request

  1. 01No card

    Create an account

    A username, a password and a code we email you.

  2. 02From $5

    Add credit

    Pay an invoice in Bitcoin. Credit lands once the payment confirms.

  3. 03Shown once

    Create a key

    One per app, each with its own spend limit.

  4. 04One line

    Change the base URL

    The rest of your code stays the same.

OpenAI SDKs
https://api.lunomi.io/v1
Anthropic SDKs
https://api.lunomi.io
Model
deepseek-v4.1-flash

Private by default

Each request leaves a billing line, not a transcript.

Time
14:02:31
Model
deepseek-v4.1-flash
Key
my-app
Tokens in
1,204
Tokens out
312
Cost
$0.0005
Prompt
Not logged
Reply
Not logged
One request, as we keep it, for 90 days. Prompts and replies aren't logged or used for training. Privacy policy
Prompt cache
Separate for each account, so nobody can learn what you sent from their own cache hits.
Keys
One per app, each with its own spend limit. Revoke one without touching the rest.
Sign-in
A password plus a code emailed each time. Every sign-in and change is in your security log.
Payment
Prepaid in Bitcoin. No card on file, nothing recurring, and spending stops at $0.
Formats
OpenAI chat completions and Anthropic messages, on one key and one balance.

Swap the base URL and keep building

Prepaid credit that never expires.

Get an API key Read the docs