Fine-tune Nemotron.Earn every time it's called.

Fine-tune Nemotron with LoRA on your own data and publish it to the store as an OpenAI-compatible API. No GPUs to set up. Each time someone calls your model, the tokens it serves become your earnings.

Build a model
2.19
Base Nemotron
4.96
Trained demo model
24 / 26
Questions won

Wednesday open?

Base Nemotron · 1 / 5

We are open Wednesday, from 11 AM to 3 PM!

Trained demo model · 5 / 5

No, the bakery is closed every Wednesday, including public holidays.

Five steps

  1. 1

    Train

    Upload a chat-format JSONL, pick a Nemotron base model and start training. You pay for the GPU time the run actually used.

    Only have documents? Nemotron drafts training examples from them.

  2. 2

    Deploy

    Your LoRA is loaded onto the base model in a vLLM endpoint. It stops on its own after 15 minutes without calls.

  3. 3

    Test

    Check real responses in the in-browser playground.

    Nemotron sends the same prompts to the base model and to your fine-tuned model, and compares the answers.

  4. 4

    Publish

    Set prices per 1M input and output tokens and per request, then publish to the store.

  5. 5

    Earn

    Users call your model at /v1 with their own API key. Each call credits 80% of your price to your earnings.

Sell a model without running GPUs

Payment split
80%
20%
YouForkline fee
No GPUs to manage
Training and inference run on Nebius serverless GPUs. Forkline reserves the machines and starts and stops the containers.
Pay only while it runs
Endpoints are billed per minute while running and stop after 15 minutes without calls. A stopped endpoint costs nothing.
OpenAI-compatible API
Users change the base_url in any OpenAI SDK and call your model. Streaming is supported.
Stripe Connect payouts
Earnings add up and can be paid out through Stripe Connect from ¥1,000.

Built on Nebius and NVIDIA

RoleNebius serviceWhat it does
LoRA trainingServerless AI JobsA PEFT + TRL container trains the LoRA on one GPU.
InferenceServerless AI EndpointsvLLM loads the base model with your LoRA and answers calls. Stops when idle.
StorageObject StorageHolds datasets and trained LoRA adapters (S3-compatible).
Data and evaluationToken FactoryCalls Nemotron as-is to draft training data from your documents and to evaluate the fine-tuned model.

Base models

  • Nemotron 3 Nano 4BDefault

    nvidia/NVIDIA-Nemotron-3-Nano-4B-BF16

    Train
    NVIDIA L40S
    Serve
    NVIDIA L40S
  • Nemotron 3.5 Lightning 30B-A3B

    nvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16

    Train
    NVIDIA H200
    Serve
    NVIDIA H100

Works with the OpenAI SDK

Point base_url at Forkline and set model to the model ID shown in the store.

API docs
Request
from openai import OpenAI

client = OpenAI(base_url="https://forkline.ai/v1", api_key="fk_...")

res = client.chat.completions.create(
    model="your-model",
    messages=[{"role": "user", "content": "How do I return an item?"}],
)
print(res.choices[0].message.content)
Response
HTTP/1.1 200 OK
X-Forkline-Cost-JPY: …

{"model": "your-model", "choices": [ … ], "usage": { … }}

Published models

All models

FAQ

Which models can I fine-tune?

Nemotron 3 Nano 4B / Nemotron 3.5 Lightning 30B-A3B. The default is Nemotron 3 Nano 4B.

Who pays for GPU time, and when?

The person who builds the model. For training, the estimate is held up front and settled on the actual time. Endpoints are billed per minute while they run.

What happens to endpoints nobody uses?

They stop after 15 minutes without calls, and no GPU time is billed. A call to a stopped endpoint starts it and returns 503 with Retry-After.

How do users pay?

Users buy prepaid credits, and each call deducts the model's price. Each user has their own API keys and balance.

Does it work with the OpenAI SDK?

Yes. Set base_url to https://forkline.ai/v1 and model to the model ID. Streaming is supported.

Is my data used to train Forkline's own models?

No. Datasets and trained LoRA adapters are kept in Nebius Object Storage until you delete the model or your account.

Publish a Nemotron trained on your data

https://forkline.ai/v1

Open the console