Build useful projects.
Choose useful rewards.
Get rewarded for the projects you are already building. Earn recognition for real results, not just signing up.
AI model API credits
$20-$10,000Potential budget-limited API access, in USD
Use your reward to build, test and ship with model families like these:
GPT
Claude
Gemini
Current partner-reported models: GPT-6 Astra, Claude Fable 5.1 and Gemini 4 Argon. Availability is confirmed for each approved offer. We aim to switch to successor models after their API release when partner support, provider terms and similar pricing allow; this is not a guaranteed release-day upgrade.
Significant inference-provider pricing or terms changes can change the available models or make API grants temporarily unavailable. Approved unspent points remain recorded and may be redeemed for other eligible available rewards under the agreed terms. Points are not cash; alternative inventory and immediate redemption are not guaranteed. Existing agreed rewards are not cancelled by a catalog update.
Model names identify intended access, not official sponsorships. Exact models, amounts and availability are confirmed with each approved offer.
Your milestone. Your choice.
After your milestone and claim are approved, choose among the eligible rewards available to your project or cohort.
The $20-$10,000 range describes possible API allowances. It is not an automatic $10,000 grant for every builder.
Read the reward termsA little room to learn, too.
DeepLearning.AI: one month. One limited offer available until claimed. Exact product and redemption terms are being confirmed; it cannot be claimed yet.
Early progress goes further.
Earn more points per eligible star at the early milestone levels, with a gradually lower rate as your project grows. Points unlock only at a reached milestone, after manual review.
| Milestone level | Eligible stars | Total points unlocked | Additional points |
|---|---|---|---|
| 1 | 20 | 800 | +800 |
| 2 | 50 | 1,700 | +900 |
| 3 | 100 | 3,000 | +1,300 |
| 4 | 250 | 6,000 | +3,000 |
| 5 | 500 | 10,000 | +4,000 |
| 6 | 1,000 | 12,500 | +2,500 |
| 7 | 2,500 | 17,500 | +5,000 |
| 8 | 5,000 | 22,500 | +5,000 |
20 stars unlock 800 points. At 7 or 19 eligible stars, no star award is claimable. Between milestones, progress continues but no additional points unlock. If 800 points were already credited at 20 stars, reaching 50 unlocks another 900 points, for 1,700 total. Each star counts once; skipping a claim does not lose the earlier eligible award. No further star levels are announced above 5,000.
These are reward points, not model tokens, cash or a transferable balance. This table applies to eligible new growth from your agreed project baseline, not an automatic conversion of every repository's displayed star count. Historical stars need a separately agreed baseline and award. Your eligible track, measurement start, evidence, funded maximum and offer terms must be agreed before reward-bearing enrollment. Existing agreed rewards are not reduced or repriced automatically. A move from an earlier schedule needs an explicit project agreement and reconciliation of points already credited.
Download and adoption milestones are also available. See the approved monthly-download and meaningful MAU/DAU tables. These are one-time high-water awards, not recurring monthly payments. Existing agreed project schedules stay valid unless a migration is explicitly agreed.
For transferred projects, eligible new downloads and usage count immediately from the agreed measurement start; the first claim is available after one month, subject to review. Existing history is not automatically treated as new activity. No duplicate awards across copied repositories, collaborators, repeated unstar/restar or overlapping counters. Bots, purchased engagement and self/team farming do not qualify.
Choose an available reward
Points can be exchanged for eligible available rewards after claim approval. The current API exchange is 10 points = $1 in non-cash API allowance; it is not a cash balance or a promise that every future reward uses the same exchange. The available reward catalog and offer terms apply.
How to claim
Milestone claims are open for manual review. Send your organization repository URL, the milestone reached, dated evidence and previous claims to hello@openroot.dev. OpenRoot verifies eligibility and points not already credited, confirms the available reward and its terms, then arranges redemption after approval. The first approved star milestone can be redeemed without waiting for another level, subject to any agreed transfer waiting period. There is no automated portal wallet or self-service redemption yet. Never email an API key or account password.
Reward useful distribution and adoption.
Usage milestone pilot, 18 October 2026. These are one-time cumulative awards per approved project and track, not paid again each month. Only reached, manually reviewed levels unlock points.
Downloads in a completed calendar month
| Level | Monthly downloads | Total points unlocked | Additional points |
|---|---|---|---|
| 1 | 100 | 200 | +200 |
| 2 | 500 | 600 | +400 |
| 3 | 1,000 | 1,000 | +400 |
| 4 | 5,000 | 3,500 | +2,500 |
| 5 | 10,000 | 5,500 | +2,000 |
| 6 | 50,000 | 10,000 | +4,500 |
| 7 | 100,000 | 12,500 | +2,500 |
100 eligible downloads unlock 200 points. After those 200 points were credited, a completed month with 500 eligible downloads unlocks another 400 points, for 600 total. Returning to the same level in a later month does not pay again.
Agree the authoritative source, package/release scope, calendar timezone, baseline, exclusions and funded maximum before accrual. Raw registry downloads are not unique people. Exclude CI, mirrors, bots, tests, repeated self-downloads and purchased activity. Where the source cannot distinguish relevant abuse, require corroboration or use an adoption track instead; not every registry exposes sufficient anti-bot data.
Meaningful product users
| Level | Meaningful MAU | Or average DAU | Total points unlocked | Additional points |
|---|---|---|---|---|
| 1 | 50 | 10 | 1,000 | +1,000 |
| 2 | 100 | 25 | 2,000 | +1,000 |
| 3 | 250 | 50 | 4,500 | +2,500 |
| 4 | 500 | 100 | 7,500 | +3,000 |
| 5 | 1,000 | 200 | 11,000 | +3,500 |
| 6 | 2,500 | 500 | 16,000 | +5,000 |
| 7 | 5,000 | 1,000 | 21,000 | +5,000 |
MAU means distinct people performing an agreed useful core product action during a completed calendar month, not visitors or signups. Average DAU is the mean of each day's distinct meaningful active users across that month, not the highest day. Do not round up to reach a level. MAU is the default recommendation; average DAU is an alternative for daily-use products.
Both full consecutive months must reach the claimed level using the same agreed measure and definition. For example, 75 and 95 meaningful MAU qualify for the 50-MAU level, not the 100-MAU level. MAU and DAU are alternatives, not two awards for the same users. Choose overlapping distribution/adoption tracks in advance so the same underlying usage is not paid twice. GitHub views and clones do not establish product DAU/MAU.
Keep the evidence useful and private.
Submit project-scoped aggregate evidence with the builder's permission and appropriate end-user notices, consent where required and access controls. Do not send account emails or raw visitor histories to OpenRoot. An analytics connection alone is not verified activity or reward eligibility.
Higher verified levels release only points not already credited for that approved project and track. A later fall in usage creates no negative points and no repeat payout. Never count the same evidence across repository copies or collaborators. New numerical levels above these tables are not announced.
Transferred projects: counting begins immediately from the agreed measurement start, with no pre-start history treated as new activity. The existing first-claim wait is one month. New adoption agreements additionally require two consecutive full calendar months of verification, so adoption claims may take longer. Existing agreed rewards and waiting periods are not changed retroactively. Moving an old proportional download rate or project-specific usage agreement onto this pilot requires an explicit migration agreement and reconciliation of prior points.
Manual claims remain open at hello@openroot.dev. Reaching a counter alone does not issue points; project eligibility, agreed baselines, funding, evidence and available offer terms still apply. There is no automated portal wallet or self-service redemption.
Real projects. Real progress.
Agree on a milestone that fits your project. Submit evidence for manual review when you reach it.
- People notice it
- Organic GitHub stars, meaningful interest or independent coverage.
- People try it
- Eligible release and package downloads in an agreed window; clones and pulls are not automatically rewardable.
- People use it
- Daily or monthly active users (DAU/MAU) performing an agreed meaningful product action. Project-scoped aggregate analytics support review; bots and tests do not count.
- It keeps delivering
- Useful adoption sustained over time, not a one-day spike.
Metrics, baseline, target, measurement window and reward are agreed before enrollment. Purchased engagement and artificial usage do not qualify. Existing projects are welcome for review; past stars or users do not automatically earn a reward.
Keep building after the hackathon.
Hackathon projects are welcome. Additional rewards for strong results in eligible, pre-approved hackathons are planned. Events, evidence, funding and bonus terms must be published and agreed before an offer begins. A win does not automatically earn points or a reward; no specific eligible events or bonuses are announced yet. Hackathon bonus claims are not open.
Bring your preferred tools.
Use a configurable client with the right endpoint, credentials and model ID. Compatibility with the actual reward gateway has not yet been tested.
- Hermes · OpenClaw · OpenCode
- These clients support custom providers or OpenAI-compatible endpoints. Gateway protocol, streaming and tool support still need verification.
- Cursor
- Custom-key chat depends on the model, client and gateway configuration. API credits do not include a Cursor subscription or Tab access.
- Codex
- A custom gateway must support the Responses API, including streaming and tool-call conversations. Chat Completions compatibility alone is not enough.
- Claude Code
- Requires a supported Claude Code gateway API format, such as Anthropic Messages. A generic OpenAI-compatible endpoint alone does not establish support.
We will confirm the setup and limitations for each offer before redemption. Never share keys publicly or paste them into an application. Codex gateway requirements · Claude Code gateway requirements
More ways to keep building.
A broader reward selection, including paid subscriptions and developer tools, is coming soon. Specific products, quantities and dates will be announced when confirmed.
Free promotion is part of the program, not something you spend a reward on. OpenRoot helps share member projects on socials and in the newsletter, completely free of charge and as much as it can. Coverage, reach and project success are not guaranteed.
Any project category, programming language or documentation language is welcome. Bring a new idea or an existing project for review. You retain your code rights and project direction; repositories are hosted in the OpenRoot GitHub organization under agreed terms.