Three open-source products past $1k/month, with three different payers

The useful question is not whether open source can make money. It is who pays, what they are buying, and which product change made the payment feel reasonable.

This week’s three cases are unusually clean on the revenue side. Verboo Code is a terminal CLI with payment-verified Brazilian subscription revenue. Rybbit is a privacy-first analytics project with a paid hosted layer and a large self-hosted audience. LLM Gateway is an AGPL gateway whose hosted credits, chat plans, coding plans, and enterprise features sit beside the open core.

The figures are snapshots, not audited financial statements. TrustMRR says its revenue pages are verified through payment-provider APIs; ListaMRR says its figures are verified through payment integrations. Currencies are left in the source currency. The common lesson is narrower than “open source works”: each project made the paid boundary specific.

What it is: An open-source coding-agent CLI for the terminal. The public repository describes it as the CLI for the Verboo platform, and the official site says the CLI is open source while the server-side models and endpoint are not. GitHub repository | official plans

GitHub stars: 14, at the time checked. Current revenue: R$30,235.90 MRR, R$71,159.71 total revenue, 166 active subscribers. The ListaMRR page was updated on July 23, 2026 and identifies the numbers as payment-API verified. ListaMRR listing

Model: Hosted version, sold as a recurring subscription for unlimited tokens, dedicated GPU-backed models, and predictable access. The code is free to inspect and install, but the service that makes the CLI valuable is paid.

The interesting part is not the star count. Fourteen stars is a warning against using GitHub popularity as a proxy for commercial demand. The paid product is a narrow operational promise: developers can run a familiar coding-agent interface in the terminal without managing external provider keys, while the service absorbs the model and infrastructure bill.

The public milestone sequence gives a more useful signal than the repository page. On July 4, ListaMRR reported R$21,550.90 MRR, R$41,211.50 in total revenue, and 133 active subscribers. On July 6 in the display timezone used here, the same tracker reported R$29,193.90 MRR and R$50,428.35 total revenue. The founder also announced the opening of the Ultra plan on July 4. R$20k MRR update | Ultra plan announcement | R$50k milestone

That is the clearest trigger in this issue: the move from a single low-friction CLI subscription to a higher-priced plan with a differentiated model bundle. The current site lists Junior at R$75/month, Pro at R$174/month, Ultra at R$900/month, and Growth API at R$600/month. Pro has a one-day trial; Junior and Ultra do not. The pricing is doing qualification work. It gives a solo developer an entry point, while making room for a high-value user who cares more about throughput and model access than about a low monthly bill.

The constraint is just as important. Verboo is not selling “open source” as the product. It is selling a managed inference service attached to an open client. That means gross margin, provider costs, abuse controls, and model availability matter more than GitHub conversion. A maintainer copying this model needs a usage boundary or infrastructure advantage. A generic CLI with no scarce backend will not inherit the economics.

Would this work for you? If you maintain a CLI that already sits in a daily workflow and users are paying someone else for a hosted API, possibly. If your tool is mostly a local utility with no expensive or inconvenient service behind it, a subscription will be hard to justify. The transferable move is not “add an Ultra tier.” It is identifying the recurring operational burden that users would rather outsource.

What it is: An open-source, cookieless Google Analytics replacement with a self-hosted edition and an EU-hosted cloud service. GitHub repository | official site

GitHub stars: 12.5k. Current revenue: $8,273 MRR, $72,703 all-time revenue, and 267 active subscriptions. TrustMRR reports $10,017 in revenue over the last 30 days, verified through a Stripe API key, with the page updated July 23, 2026. TrustMRR revenue page

Model: Hosted version. The source is available under AGPL-3.0, while the paid offer removes the infrastructure work: setup, updates, scaling, backups, and a managed analytics environment. Current public pricing is $13/month for Standard, $26/month for Pro, and custom pricing for Enterprise. GitHub metadata | Rybbit pricing and product page

Rybbit is the more familiar open-core story, but the numbers explain why the boundary works. Analytics is not a one-time utility once a team has historical data, dashboards, retention requirements, and multiple people reading reports. A free self-hosted option creates distribution and trust. The cloud plan charges for the part that becomes operationally annoying after adoption.

The recent trigger is less clean than Verboo’s, and that is worth saying plainly. Rybbit’s public terms, dated November 8, 2025, describe both cloud-hosted and self-hosted software. Its current site makes that split concrete with managed cloud pricing, a seven-day trial, and enterprise options including isolated instances and on-premise installation. terms and conditions The record supports a product shift toward a hosted, paid layer within the channel’s lookback window. It does not isolate the exact month when Rybbit crossed $1k MRR or prove that one feature caused the current run-rate.

That limitation does not make the case useless. It changes the conclusion. The public evidence supports “hosting became the monetizable boundary,” not “session replay caused $8,273 of MRR.” The repository’s latest release was v2.7.0 on July 7, 2026, and the project continues to improve the free product. That is a reminder that the open core still has to earn distribution; the cloud layer cannot compensate for a product that stops being useful when self-hosted.

There is also a scale lesson. Rybbit has 12.5k stars and about 10,000 users according to the TrustMRR page, but 267 active subscriptions. Stars are not the funnel. They are one top-of-funnel indicator among many. The paid conversion comes from a narrower cohort that values privacy, compliance, and not operating ClickHouse-backed analytics themselves.

Would this work for you? If your project stores state, needs upgrades, or has a painful deployment path, hosted operations are a strong candidate. If self-hosting is trivial and the project has no ongoing data, uptime, or compliance burden, charging for hosting may produce little willingness to pay. The test is whether the cloud plan removes a repeated operational task, not whether the repository has enough stars.

What it is: An open-source API gateway that routes requests across LLM providers, tracks cost and usage, and exposes one interface for many models. The public repository is AGPLv3, with commercial enterprise features in the ee/ directory. GitHub repository | official site

GitHub stars: 1.5k. Current revenue: $43,144 MRR, $654,362 all-time revenue, and 776 active subscriptions. TrustMRR reports $143,101 in the last 30 days, verified through Stripe, with the page updated July 23, 2026. TrustMRR revenue page

Model: Hosted version with usage-based credits, subscriptions, and an enterprise upsell. Self-hosting is free under AGPLv3. The hosted side charges a 5% platform fee on credits, while Chat and DevPass add recurring plans. pricing and product page

This is not just “OpenRouter, but open source.” The monetization surface expanded in public steps. The founders’ June 11, 2026 build-in-public article says they started building in May 2025 and reached a $470k revenue milestone in twelve months. founder milestone article The official blog records an April 12, 2025 launch, then an October 2025 Chat product, credits promotions in November and December, and a Q2 2026 expansion into coding plans, subscriptions, and enterprise controls. official blog

The most concrete recent trigger was packaging. On June 9, the co-founder announced Chat plans with Starter, Plus, and Pro subscriptions alongside pay-as-you-go credits. On June 16, LLM Gateway Chat launched on Product Hunt with 210-plus models and subscriptions from $9 to $49/month. On July 1, the Q2 roundup described DevPass coding plans, Chat subscriptions, enterprise IAM, and 27 million routed requests. Chat plans | Product Hunt launch | Q2 roundup

The lesson for maintainers is about revenue surfaces, not AI fashion. The open gateway attracts technical users. Credits monetize variable usage. Chat plans monetize a repeatable end-user workflow. DevPass monetizes a coding-specific workload. Enterprise controls monetize procurement and governance. One codebase can support several payers when each offer maps to a distinct job.

The risk is complexity. More plans mean more billing states, support obligations, and cost exposure. An LLM gateway also pays provider bills before it collects its own margin. A maintainer without reliable metering, rate limits, and a clear provider-cost pass-through should not copy the plan names. Copy the separation between open routing infrastructure and paid convenience, not the surface area.

Would this work for you? If your project sits in a spend path and can measure usage, a credit or metered model may fit. If users cannot predict the value of another usage bill, a simpler annual license or support contract may be healthier. The model works here because the gateway controls routing, billing, and visibility in one place.

The payer is different in each case. Verboo sells predictable access to an expensive backend. Rybbit sells the removal of deployment and compliance work. LLM Gateway sells aggregation, metering, and multiple paid workflows around a provider-heavy infrastructure layer.

The common trigger is a paid boundary that became more specific. Verboo introduced a higher-priced model tier. Rybbit made managed hosting a first-class product alongside self-hosting. LLM Gateway added credits and subscriptions around a gateway that can still be run independently. None of these cases was unlocked by a sponsorship button alone.

The evidence also argues against a simple stars-to-revenue formula. Verboo has 14 stars and a reported 166 subscribers. Rybbit has 12.5k stars and 267 subscriptions. LLM Gateway has 1.5k stars and 776 subscriptions. Different projects convert different kinds of trust. Repository visibility helps, but the payment boundary determines whether visibility becomes income.

The practical filter is simple: name the recurring burden, name the payer, and prove the change that made the paid offer easier to accept. If those three answers are vague, the monetization model is probably still a hope rather than a business.