LLM Gateway passed $1 million in cumulative Stripe collections a little more than a year after launch. TrustMRR also recorded $78,764 in monthly recurring revenue and 1,386 active subscriptions.
Those numbers show that developers will pay for a simpler way to work with a fragmented AI market. They don't show how much of that money LLM Gateway keeps.
The distinction matters because the company sells both software and access to third-party models. A customer can buy hosted credits that ultimately pay OpenAI, Anthropic, Google, or another provider. The full payment may pass through Stripe even though only a fraction belongs to the gateway.
That makes LLM Gateway more interesting than a straightforward seven-figure SaaS story. Luca Steeb and Ismail Ghallou simplified the work of switching between AI models, then used open source, search, and free access to get the product in front of developers. The next test is whether that position becomes a valuable control layer or remains primarily a routing layer built on third-party model infrastructure.
LLM Gateway Revenue at a Glance
The public figures come from different dates and describe payment activity rather than audited accounts. TrustMRR's final snapshot is useful, but its connected Stripe key had expired by September 2026.
What the Public Record Actually Shows
Three measures from Stripe-connected records, each answering a different question.
Source: TrustMRR, from Stripe-connected records. Launched April 2025. These are payment records rather than audited accounts, and subscriptions are not the same as customers.
The safest conclusion is that LLM Gateway processed more than $1 million through its connected Stripe account. Public data does not reveal provider payments, gross profit, customer concentration, or the mix between hosted credits and higher-margin software fees.
Who Founded LLM Gateway and Why They Built It
Steeb and Ghallou, who publishes online as Smakosh, introduced LLM Gateway in April 2025, as developers were gaining access to more capable models from a growing number of providers. Each new provider brought another set of credentials, rate limits, prices, dashboards, and response formats.
A team could prefer one model for coding, another for reasoning, and a cheaper option for routine requests. Using all three meant maintaining separate integrations and rewriting parts of the application whenever a provider changed.
LLM Gateway placed one stable interface above that churn. Its OpenAI-compatible API let developers change a base URL instead of rebuilding each integration. The gateway handled provider selection and returned a standardized response.
The first promise was practical: model choice without integration sprawl.
Open Source Turned Trust Into Distribution
Asking developers to route production traffic through a new company requires trust. LLM Gateway addressed that problem by making its core product inspectable and self-hostable.
The public GitHub repository uses an AGPLv3 license for the core gateway, while commercial capabilities sit under a separate enterprise license.
Open source gave technical users a way to examine the product before committing traffic or money. It also created an acquisition path that did not depend on a large sales team.
Search captured developers who already wanted an alternative
Developers were already looking for ways to manage multiple AI providers. LLM Gateway focused on appearing when they searched for a solution.
Ghallou said a Reddit post describing LLM Gateway as a self-hosted OpenRouter alternative helped the company rank for that search intent. He later said organic search accounted for roughly 85% of LLM Gateway's website traffic. Both figures are founder-reported, but the site reflects the strategy: competitor alternatives, model comparisons, integrations, migration guides, and token-cost pages all address existing technical demand.
Free access created more than one paid path
Free access gave developers a way into the product before LLM Gateway asked them to pay.
Four Ways In, Two of Them Paid
Adoption comes first, procurement later. Each step raises the operational burden the gateway removes.
Source: LLM Gateway's own pricing and documentation. The company has not disclosed how revenue splits across these paths.
This sequence allowed adoption to come before procurement. A developer could test the routing layer first, then pay when the operational burden or security requirements grew.
The First Year Turned Technical Interest Into Payment Activity
Ghallou documented the first 12 months in an X retrospective. He reported 34.4 million API requests, 214 billion routed tokens, 78,000 visitors, 8,800 signups, 714 paying customers, and $470,000 processed.
The First Twelve Months, as the Founder Tells It
From open-source launch to paid usage. These are company figures covering an earlier period than the TrustMRR snapshot.
The first three stages are the founder's own account; the final figure is TrustMRR's Stripe-connected record. None of it is audited.
Those are company-reported figures, and they cover an earlier period than the TrustMRR snapshot. More importantly, LLM Gateway was becoming part of recurring application traffic. Once a gateway sits inside a working application, switching away takes more effort than simply trying a competing product.
The Second Half of the Million Came Faster
The pace after the first anniversary is the clearest sign that LLM Gateway's growth was still accelerating. Ghallou's June 2026 retrospective put first-year processed revenue above $470,000. TrustMRR recorded $1,000,116 by September 7. In those public snapshots, the gateway added more than $530,000 in cumulative collections in less than three months, exceeding everything it processed during its first year.
First Year Against the Three Months That Followed
Cumulative Stripe collections at two public snapshots. The second period is shorter and larger.
Sources: the founder's June 2026 retrospective and TrustMRR's September 7, 2026 snapshot. The second figure is the difference between the two, not a separately reported number.
The same retrospective reported 8,800 signups and 714 paying customers. That earlier figure is not directly comparable with TrustMRR's later count of 1,386 active subscriptions. At the first-year snapshot, the reported figures worked out to roughly eight paying customers for every 100 signups, although this should not be treated as a formal cohort conversion rate. The 34.4 million API requests and 214 billion tokens also show why a gateway can expand without adding a new customer for every dollar: existing applications keep generating traffic after integration.
The model gives LLM Gateway several ways to turn developer interest into revenue. Search and open source bring developers in, while free bring-your-own-key access lets them test the interface without paying LLM Gateway. Hosted credits add a paid convenience layer, and enterprise features give larger teams another reason to spend.
What LLM Gateway Does for AI Development Teams
LLM Gateway starts with a simple proposition: access many AI models through one API. But sitting between applications and model providers also gives it a role in routing, monitoring, cost control, and reliability.
The platform can route requests across providers and models, fail over when a provider becomes unavailable, track tokens, latency, errors and spending, centralize provider credentials, and support both hosted and self-managed deployment.
As more application workloads move onto cloud infrastructure, cloud computing statistics show how deeply cloud services have become embedded in business IT. For LLM Gateway, basic routing is only part of the opportunity. The company can add more value by helping teams decide which model handles each request, enforce policies, manage costs, and keep applications running when a provider fails.
That is the difference between a convenient adapter and an operational control plane. The first saves integration time. The second becomes part of how a company governs its AI systems.
The $1 Million Figure Is Payment Volume, Not Proven SaaS Revenue
LLM Gateway's pricing helps explain why the $1 million figure needs context. Bring-your-own-key access and AGPLv3 self-hosting are free. Hosted credits carry a 5% platform fee on credit purchases, according to the company's pricing page.
Where a $100 Credit Purchase Goes
A simplified illustration, not disclosed figures. It shows why collections and software revenue are different measures.
Illustrative only. LLM Gateway has not published the split between model credits, software fees and enterprise contracts, nor its gross margin or provider costs.
That does not make the collections meaningless. Higher payment volume still matters because it means more customer usage is flowing through LLM Gateway, creating opportunities to sell governance, support, and enterprise deployment. It does mean cumulative collections should not be described as $1 million in ARR.
LLM Gateway has not disclosed the split between model credits, software fees, and enterprise contracts. It has also not published gross margin or provider costs. Until it does, the Stripe milestone proves demand and payment activity, not the economics behind them.
LLM Gateway Competes With Both Startups and Infrastructure Platforms
LLM Gateway competes with specialist gateways, open-source projects, and infrastructure companies that can bundle similar features into products developers already use.
OpenRouter offers broad model access and a large network. LiteLLM is a widely adopted open-source proxy. Portkey, Helicone, Requesty, and Braintrust combine routing with observability, evaluation, or governance.
Some rivals also have far more capital. TechCrunch reported that OpenRouter raised $113 million at a $1.3 billion valuation. This research found no credible institutional funding round for LLM Gateway. That does not prove the company is bootstrapped, but it does mean no comparable capital base has been publicly disclosed.
Cloudflare, Vercel, Kong, and major cloud providers create pressure from another direction. They can place routing, security, and observability inside infrastructure customers already buy, as cloud computing trends continue to push more application infrastructure toward cloud platforms.
Model access alone is unlikely to differentiate LLM Gateway for long because competitors can connect to many of the same providers. Its differentiation will have to come from areas such as deployment, provider coverage, controls, traffic data, and developer distribution.
Growth Now Depends on Owning the Control Plane
LLM Gateway grew because it removed a problem developers could feel immediately. One interface was easier than a growing collection of provider integrations. Open source made the product easier to trust, while search placed it in front of users already looking for that answer.
The $1 million Stripe milestone shows that the system moved beyond experimentation. Customers were willing to route real spending and recurring workloads through it.
But the same architecture that produced the headline also complicates it. Hosted credits can make collections grow quickly while much of the money flows upstream. Free usage can expand adoption without creating direct revenue. Enterprise controls may offer better margins, but their contribution is still unknown.
The opportunity is to become the neutral layer companies rely on to choose models, enforce policies, and control costs. If LLM Gateway remains primarily a payment and routing layer, it may have less room to capture the value flowing through its platform.
Crossing $1 million proves that developers wanted the gateway. How much of that value LLM Gateway can keep will determine whether it owns the control plane or merely carries the traffic. For another company whose headline number needed the same kind of unpacking, see our investigation into Cometly's $251,000 launch week.
