The Lepton AI free plan (now NVIDIA DGX Cloud Lepton) gives you real compute capacity for $0/month: up to 48 CPUs and 2 GPUs running at the same time, plus 1 GB of storage and 10 GB/month of network egress included.
Solo devs testing an OpenAI-compatible API, startup teams spinning up GPU endpoints without committing to a contract, researchers prototyping a model service quickly. This plan can cover all of that, as long as you understand the limits.
This guide breaks down Lepton AI free credits and limits, how to get Lepton AI credits via the Basic plan, who qualifies, and the practical ways to stretch it before you pay anything.
Program at a Glance
| Provider | Lepton AI (NVIDIA DGX Cloud Lepton) |
| Credit Amount | $0/month plan + 1 GB storage + 10 GB egress |
| Duration | Ongoing (Basic plan stays free while offered) |
| Eligibility | Anyone who creates a Basic Plan account |
| Credit Card Required? | No upfront payment; usage-based billing applies |
| Difficulty | Easy; signup and start using immediately |
| Best For | Prototyping, small endpoints, GPU experiments |
| Official Page | Lepton AI Program Page |
What You Actually Get
The Basic (Free) plan is a $0/month entry point to DGX Cloud Lepton. In one single-user workspace, you can run up to 48 CPUs and 2 GPUs concurrently. You also get a small free allowance for storage (the first 1 GB) and network egress (the first 10 GB per month). On top of that, Lepton offers serverless endpoints for popular open-source models with an OpenAI-compatible API, plus a Pythonic SDK and CLI that let you deploy custom services without needing Docker or Kubernetes.
In real terms, this is enough to prove out an idea. You can stand up a serverless LLM endpoint for a demo, run a Dev Pod for interactive work (Jupyter/SSH/VS Code), or deploy a custom “Photon” service for a small internal tool. The tight part is not “can it run” but “how much traffic can it handle,” because the free plan rate limit for serverless endpoints is only 10 requests per minute.
Who Qualifies (and Who Doesn’t)
Eligibility is refreshingly simple: if you can create an account, you can use the Basic plan. There is no application process, no accelerator requirement, and no “approved startup” gatekeeping called out for the free tier.
- You need to create a Lepton account using email or a supported social login (GitHub or Google).
- The Basic plan is limited to a single-user workspace, so it is not designed for teams sharing one workspace.
- You must stay within the Basic plan caps: 48 concurrent CPUs, 2 concurrent GPUs, and 10 requests per minute on serverless endpoints.
- Usage is pay-as-you-go once you go beyond free storage/egress or start consuming paid resources.
If you need multi-user workspaces or higher serverless rate limits, the Basic plan won’t fit. Also, if you assumed “free” meant unlimited serverless calls or free GPU hours, that’s not what this is.
How to Sign Up
Registration is quick, but you will want a terminal handy if you plan to deploy via the SDK.
- Go to lepton.ai (it redirects to NVIDIA’s DGX Cloud Lepton page).
- Click “Get Started” or “Sign Up.”
- Create an account using your email or a social login (GitHub, Google).
- After signing in, open the Lepton Dashboard and create a workspace.
- Install the Python SDK locally:
pip install -U leptonai. - Authenticate your CLI with:
lep login, then follow the prompts to link your account. - You are now on the Basic Plan with no subscription fee, and you only pay for resources you actually consume.
One small gotcha: the marketing site redirects to NVIDIA’s domain, but the legacy Lepton dashboard and docs are still where you will do the actual work. Don’t overthink it.
What the Credits Cover
This “free credits” program is really a free plan with included allowances and defined caps. You get free storage and free egress up to the included amounts, and you can access the platform’s core workflows: serverless endpoints (with rate limits), dedicated GPU compute billed by the minute, Dev Pods, Batch Jobs, and custom model deployments through the SDK.
| Service / Feature | What It Does | Included? |
|---|---|---|
| Serverless LLM endpoints | Hosted open-source inference via OpenAI-compatible API | Partial |
| Dedicated GPU compute | Run custom deployments/training billed by the minute | ✓ |
| Dev Pods | Interactive development (Jupyter, SSH, VS Code) | ✓ |
| Storage & network egress | Keep artifacts and move data out of the platform | Partial |
Notable exclusions: the Basic plan does not give you multi-user workspaces, and serverless endpoints are constrained by a 10 RPM rate limit. Also, model availability is dynamic and depends on cloud partner routing, so you can’t assume a fixed catalog will always be there.
Limitations to Know About
Every free program has catches. With DGX Cloud Lepton Basic, the biggest ones are concurrency caps and traffic limits, not a hard time window.
- The Basic Plan workspace is single-user only, which blocks most team workflows.
- Serverless API endpoints are rate-limited to 10 requests per minute on the Basic Plan.
- Storage is only free for the first 1 GB, and then it is billed at about $0.15 per GB per month.
- Network egress is only free for the first 10 GB per month, and then it is billed at about $0.15 per GB.
What happens when you hit the free allowances or caps depends on what you are doing. There is no subscription that suddenly starts charging $30/month; Basic stays $0/month. But usage-based billing applies once you consume paid resources (compute minutes, extra storage, extra egress), so you should treat it like any other pay-as-you-go cloud account and keep an eye on usage.
Have Unused Lepton AI Credits?
It happens more than people admit. Teams pick up compute credits through partner deals or bigger programs, then the product direction changes and the credits sit there until they expire. If you are holding Lepton AI credits you won’t use, you can turn “dead value” into budget for something else by selling them instead of letting them lapse.
Need More Lepton AI Credits?
Once you outgrow the Basic plan limits (usually the 10 RPM serverless cap), paying retail is not your only option. AI Credit Mart lists discounted Lepton AI credits from orgs with surplus allocations, which can be an easy way to keep building while spending less. Discounts typically land around 30–70% below face value, depending on what’s available.
Tips for Getting the Most Out of Your Credits
- Start on Basic and keep it that way until you are sure you need multi-user workspaces or higher RPM, because the free plan has no subscription fee at all.
- If you want “drop-in” compatibility, use the OpenAI client libraries and only change the base_url to
https://<model-name>.lepton.run/api/v1/plus your Lepton API key. - Keep experiments small and artifacts lean so you stay under the first 1 GB of free storage.
- Plan around the 10 requests per minute limit for serverless endpoints; frankly, it’s best treated as a prototyping throttle, not a production quota.
- Use Python 3.10+ with the
leptonaiSDK, because that is the recommended baseline for smoother local dev and deployments.
Frequently Asked Questions
There’s no lump-sum dollar credit; the value is the $0/month Basic plan plus included allowances (1 GB storage and 10 GB/month egress) and the ability to use serverless endpoints and run up to 48 CPUs and 2 GPUs concurrently in one workspace.
No upfront payment is required; it’s usage-based billing, so you only pay if you consume paid resources.
The Basic plan is ongoing (no stated expiration), but the platform is in transition post-acquisition, so plan structures and pricing may change over time.
Yes. If you have Lepton AI credits you won’t use before they expire, you can list them on AI Credit Mart and sell them at up to 70% of face value. Companies regularly list surplus credits from startup programs and enterprise agreements.
AI Credit Mart has discounted Lepton AI credits available from companies with surplus allocations. Prices are typically 30-70% below retail.
On the Basic plan, there is no credit balance to expire; instead, free allowances reset by policy (like monthly egress) and anything beyond the included amounts is billed pay-as-you-go.
10 requests per minute.
Dedicated GPU compute is billed by the minute with no minimum commitment, and the Basic plan allows up to 2 GPUs concurrently. Listed options include NVIDIA A10, RTX A6000, H100 80GB, and A100 80GB (with A100/H100 able to scale to 1, 2, 4, or 8 GPUs on supported plans). Model availability and partner routing can affect what you can access at any given time, so you will want to confirm current options in the dashboard before building around a specific SKU. If you stay idle and don’t spin resources up, you pay nothing.
The Lepton AI Basic plan is one of the easier ways to get hands-on DGX Cloud Lepton access without committing to a subscription. Use it to prototype smartly, then either upgrade or source discounted credits if you need to scale.
Your AI credits are losing value every day
Join the marketplace and start trading unused credits today.