NULLMOTH / LAB

Your token lab.

Test Recon, copy a request or estimate your usage.

01 / THE MODEL

NullMoth Recon.
Start experimenting.

Test the integration and evaluate response quality on your own tasks.

Status unchecked

NullMoth Recon

Fine-tuned for cybersecurity and red-team workflows. Review code, explore attack scenarios and build your security tools.

Try the model

Commercial access in preparation· Accounts and keys are available in this MVP. Development credits are for testing.

Input · provisional price / million: 1.95$
Output · provisional price / million: 6.50$
Local test: Streaming · up to 1,024 output tokens per request
Test context: 24,000 characters total · up to 20 messages

Estimate my usage View capabilities and limits

03 / TRY IT

Send a prompt.
See what happens.
Explore.

Ask a question, explore an idea and continue the conversation.

Your conversation space.The backend sends your messages to Recon. Without a server connection, the examples are illustrative only.

Prefer code?

PlaygroundPreview

NULLMOTH / WRITTEN EXAMPLE

A useful review starts with missed cases: empty input, unexpected values and network failures. For each case, write a test and a minimal fix. You decide which changes to apply.

Code review example · illustrative response

04 / FIRST REQUEST

Your application.
Our inference.

A familiar format for your application, assistant or next prototype.

A familiar format
Messages, roles and responses in chat completions format.
Measurable usage
Input and output token counts returned in the usage field.
An output budget per request
Set max_tokens. Test limits are documented.

Read the integration guide

Example environment

your-next-project / first-request

curl "$ATELIER_BASE_URL/chat/completions" \
  -H "Authorization: Bearer $ATELIER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "nullmoth-recon",
    "messages": [{
      "role": "user",
      "content": "Construisons quelque chose."
    }],
    "max_tokens": 512,
    "stream": false
  }'

Commercial template: set the API URL and your key when the service opens.

Concurrent requests?Understand batching.

05 / TOKENS ON YOUR TERMS

Choose your budget.
Track your usage.

Prepaid credits. Choose a budget; token usage is deducted from your balance.

Launch offer in preparation

FIRST EXPERIMENTS

Scout

5USDC

Your first requests and an idea to explore.

$5 API credit
A manageable starting budget
No subscription

FOR YOUR NEXT PROJECTBUILD MOMENTUM

Build

20USDC

For development, iteration and your agents.

$20 API credit
The same per-token rate
No subscription

GO FURTHER

Fleet

100USDC

One balance for several tools.

$100 API credit
A shared balance across your projects
No subscription

Provisional pricing. No payment is accepted in this preview. Final rates and credit terms will be confirmed before launch.

NO BLACK BOX

Usage, in numbers.

Estimate a set of requests. Input tokens include your messages and context; output tokens are the generated response.

Provisional price / million tokens
1.95$ input 6.50$ output

Estimate unit

Input(million tokens)

Output(million tokens)

Estimated costExcluding any network fees3.58 $

Estimate at the provisional real-time rate. Billing uses actual token counts.

Your credits, in USDC.Ethereum planned · payments in preparation

How it will work

06 / BEFORE YOU START

Good
questions.

Can I use the API yet?

On this machine, yes: the backend provides Recon chat completions when the inference connection is active. Commercial access is in preparation. Create an account in your account area, add test credits and create an API key. Chat uses the same balance.

Which model is available?

NullMoth Recon uses OrcaRouter’s Qwen3.8-27B Uncensored FP8 variant. Responses are not guaranteed to be accurate or to fulfill every request. You control verification and permissions for code, analysis and agent actions.

How will USDC payments work?

Ethereum is the selected launch network. Automatic payment verification is not yet implemented for Ethereum. The current local development flow uses simulated credits; the older Solana adapter remains a separate test capability.

This preview does not request a transfer. Credit expiry, account recovery and refund terms will be published before commercial launch.

Why credits instead of unlimited usage?

A short question and a long document use different resources. Input and output tokens are charged separately, and requests stop when the available balance cannot cover them.

Is a Batch API available?

Not yet. Asynchronous batch processing and discounted rates are under consideration. No public batch endpoint is currently offered.

Do I need to rebuild my tools for NullMoth?

The API targets the OpenAI-compatible chat completions format. Start with the URL, key and model ID. Check the capabilities your client needs in the integration guide.

Your token lab.

NullMoth Recon.Start experimenting.