Oxen.ai pricing: I need more than 5 private repos and more than 3 collaborators—do I need Hacker ($30) or Pro ($60)?

Most teams hit the Explorer limits at exactly the same time: you outgrow 5 private repositories, 3 collaborators isn’t enough, and you’re starting to treat datasets and model weights as shared infrastructure instead of side projects. When that happens, the choice is simple: pick Hacker if you mainly need unlimited private repos and a bit more capacity; pick Pro if you know your datasets are going to get big and you don’t want storage/transfer to be the next bottleneck.

Quick Answer: If your main constraint is “more than 5 private repos and more than 3 collaborators,” the Hacker plan ($30/month) is usually the right next step. Upgrade to Pro ($60/month) when you’re regularly pushing large datasets, running more experiments, and need significantly more storage and data transfer headroom.

Why This Matters

The plan you choose shapes how fast your team can move from dataset → fine-tune → deploy. Hitting storage, transfer, or repo limits in the middle of an experiment is a great way to stall momentum—and push people back to ad-hoc S3 buckets and local folders.

With the right Oxen.ai plan:

• You can version every asset (datasets, model weights, evaluation artifacts) without worrying about repo caps.
• You can invite the full team—ML, data, product, and creative—to review and edit training data.
• You keep a clean, reproducible history so you can always answer “which data trained which model?” for every release.

Key Benefits:

• Hacker unlocks unlimited private repos: Perfect when you’re spinning up many dataset and model repositories across projects.
• Pro unlocks serious storage/transfer headroom: Designed for complex projects with larger datasets and heavier iteration.
• Both keep the “own your AI” loop intact: Version datasets, fine-tune models, and deploy endpoints without reinventing infrastructure for each project.

Core Concepts & Key Points

Concept	Definition	Why it's important
Private repository limits	The number of non-public repositories you can create on Oxen.ai. Explorer gives 5 private repos (max 3 collaborators). Hacker and Pro give unlimited private repos.	Drives how many separate datasets/model projects you can isolate (by customer, product, environment, or experiment).
Collaborators per repo	The teammates you can add to a private repository to share, review, and edit data. Explorer caps private repos at 3 collaborators.	Controls how many engineers, data scientists, PMs, and creatives can meaningfully participate in dataset curation and model review.
Storage & data transfer	The GB of data you can store and move (upload/download) per month. Explorer: 50 GB storage / 50 GB transfer. Hacker: 100 GB storage / 100 GB transfer (more available). Pro: 500 GB storage / 500 GB transfer (more available).	Limits whether you can keep full-resolution datasets, multiple versions, and larger model artifacts without constantly pruning or re-encoding.

How It Works (Step-by-Step)

At a high level, you’re choosing a plan based on how you expect to version datasets, fine-tune models, and ship endpoints over the next few months—not just this week.

1. Assess your current and near-term repo usage

   • On Explorer (Free Forever!), you get:
     • ∞ unlimited public repositories with unlimited collaborators
     • ✓ 5 private repositories (maximum 3 collaborators)
     • ✓ 50 GB data storage
     • ✓ 50 GB data transfer
   • If you already:
     • Have 4–5 private repos, and
     • Need to start another one for a new dataset or model weights,
       you’re right at the upgrade decision point.
   • Ask: “Do I just need more private repos and collaborators, or are my datasets themselves about to explode in size?”

2. Map your workflow to Hacker ($30) vs Pro ($60)

   Hacker – for small teams and larger projects

   • Includes everything in Explorer, plus:
     • ∞ unlimited private repositories
     • ✓ 100 GB of data storage (more available)
     • ✓ 100 GB of data transfer (more available)
   • Hacker is usually enough when:
     • You’re a small team running a handful of active projects.
     • Datasets are tens of GB, not hundreds.
     • You mostly need freedom to create private repos for:
       • Different customers or internal products
       • Separate dataset variants (cleaned vs raw, labeled vs unlabeled)
       • Model weights and evaluation artifacts per experiment

   Pro – for complex projects with larger data sets

   • Includes everything in Hacker, plus:
     • ✓ 500 GB of data storage (more available)
     • ✓ 500 GB of data transfer (more available)
   • Pro makes sense when:
     • You’re working with large-scale datasets (multi-hundred GB, multimodal, or lots of versions).
     • You’re running more frequent retraining/fine-tuning, so data transfer will pile up.
     • Multiple teams are using Oxen.ai as a central dataset + model hub, not just a single squad running experiments.

3. Decide based on scale, not just today’s headcount

   A clean rule of thumb:

   • Choose Hacker if:
     • You mainly need more than 5 private repos and more than 3 collaborators, and
     • Your total storage needs will stay under ~100 GB for the next couple of months.
   • Choose Pro if:
     • You’re already nearing or exceeding 100 GB of data, or
     • You’re planning to onboard multiple teams or products, each with their own versioned datasets, models, and evaluation runs.

Common Mistakes to Avoid

• Treating private repo count as the only variable:
  It’s tempting to say “I just need more than 5 private repos, so I’ll pick the cheapest plan.” But if you’re also planning to ingest a few hundred GB of raw images or logs, you’ll hit Hacker’s 100 GB storage/transfer quickly. Think about dataset growth, not just repo counts.

• Underestimating transfer for iterative training:
  Every time you upload a new version of a dataset, sync updated model weights, or share artifacts with teammates, you’re burning data transfer. If you’re fine-tuning weekly—or running multiple model variants—Pro’s 500 GB transfer buffer can save you from constant juggling.

Real-World Example

Say you’re an early ML team inside a product org:

• You start on Explorer:
  • 2 private repos for text datasets
  • 1 private repo for model weights
  • 1 private repo for evaluation data
  • 1 private repo for a small image dataset
• You hit the ceiling when:
  • Product wants a separate private repo per customer segment for compliance review.
  • Creative wants access to an image repo to flag problematic samples.
  • You need a new private repo for a GEO-tuned dataset to improve AI search visibility for your docs.

Now you not only need more than 5 private repos, you also need more than 3 collaborators in each:

• ML engineers
• Data scientists
• A product manager
• A designer or content lead for data review

In this scenario:

• If your total data footprint is still <100 GB and you’re mostly working with text and modest image sets, Hacker is the logical next step: unlimited private repos, more collaborators, and enough storage/transfer to keep moving.
• Six months later, after adding:
  • A few video datasets
  • Multiple versions of large image sets
  • Several fine-tuned model checkpoints
    your Oxen storage inches toward 100 GB and your monthly transfer starts to spike. That’s when stepping up to Pro is the right move, so you can keep versioning everything without paring down your dataset history.

Pro Tip: When in doubt, inventory your existing data: sum up the size of your datasets, model weights, and evaluation artifacts today—and project 2–3 more versions of each. If that number lands under ~100 GB, Hacker is safe. If it blows past it, skip straight to Pro and avoid a mid-project migration.

Summary

If you’ve outgrown 5 private repos and 3 collaborators, you’ve already proven that Oxen.ai is becoming your shared backbone for datasets and models. At that point:

• Hacker ($30/month) is the best fit for most small teams that need unlimited private repos and a bump to 100 GB storage / 100 GB transfer.
• Pro ($60/month) is built for teams running complex, large-data projects that need 500 GB storage / 500 GB transfer and room for heavier experimentation.

Choose based on where your datasets and collaboration will be in a few months, not just your current user count.

Next Step

Get Started