wouldit.work

Developer tools

LLM Gateway: an AGPL LLM router that passed $1M verified revenue in about a year

LLM Gateway gives developers one API for many model providers, open source with a paid cloud. TrustMRR shows it crossed $1M in Stripe-verified revenue within roughly a year of its 2025 founding.

Founder
Luca Steeb and Smakosh
Launched
2025
Model
Open-core: free bring-your-own-key plan, 5% fee on credit usage, Enterprise tier

Published · 2 min read · 7 sources

Key takeaways

  • One API to route requests across many LLM providers; code is AGPLv3 with a commercial ee/ directory.
  • GitHub repo created Apr 2025; TrustMRR lists it as founded May 2025.
  • TrustMRR (Stripe-verified, data last synced 7 Sep 2026): $1,000,116 all-time revenue, $78,764 MRR, 1,386 active subscriptions.

What it is

LLM Gateway is an API gateway that sits between your app and model providers such as OpenAI, Anthropic and Google, so you can route requests, manage provider keys and track spend in one place (TrustMRR). The homepage calls it "the open-source LLM API gateway" and offers self-hosted or cloud use (LLM Gateway). The repository license is AGPLv3, except an ee/ directory under a commercial license (LICENSE).

Timeline

  • 12 Apr 2025: the theopenco/llmgateway repository is created (GitHub).
  • May 2025: founding date on TrustMRR, which lists Luca Steeb and Smakosh as founders (TrustMRR).
  • 7 Sep 2026: last revenue sync on TrustMRR: $1,000,116 all-time revenue, $78,764 MRR and 1,386 active subscriptions, verified via Stripe (TrustMRR).
  • 4 Nov 2026: TrustMRR notes the Stripe API key has expired, so newer figures are not shown (TrustMRR).

Why it worked

  • A real daily pain. The problem it states is juggling "multiple API keys and provider dashboards" (TrustMRR), which every team using more than one model hits.
  • Free entry, fee on usage. Bring-your-own-key is "free forever" (LLM Gateway), while buying credits through the gateway carries a 5% fee (pricing). Revenue grows with customers' AI spend.
  • Open source builds trust for infrastructure. Developers can read and self-host the code (LLM Gateway), which lowers the risk of routing all model traffic through a young company.
  • Enterprise upsell. SSO, audit logs, guardrails and an uptime SLA sit in the Enterprise tier (pricing).

What to learn

  1. Charging a small percentage on usage you already route lets revenue scale with the customer without new sales work.
  2. Open core works for infrastructure: keep the core open and put team and compliance features in a paid tier.
  3. Pick a license on purpose. AGPL plus a commercial directory is one way to stay open while protecting the paid features.

Risks and what didn't work

  • The market is crowded: other self-hosted gateways keep launching on Hacker News, such as Foreman and Mantis in mid-2026.
  • Part of the credit revenue is pass-through spending on model providers, so gross revenue overstates what the company keeps (inference from the 5% fee model on the pricing page).
  • Provider price cuts or provider-native routing could squeeze a thin margin.
  • TrustMRR figures stopped updating on 7 Sep 2026 because the Stripe key expired (TrustMRR).

FAQ

Short answers drawn from the sections above.

What is LLM Gateway?

LLM Gateway is an API gateway that sits between your app and model providers such as OpenAI, Anthropic and Google, so you can route requests, manage provider keys and track spend in one place (TrustMRR). The homepage calls it "the open-source LLM API gateway" and offers self-hosted or cloud use (LLM Gateway). The repository license is AGPLv3, except an ee/ directory under a commercial license (LICENSE).

Why did LLM Gateway work?

  • A real daily pain. The problem it states is juggling "multiple API keys and provider dashboards" (TrustMRR), which every team using more than one model hits.
  • Free entry, fee on usage. Bring-your-own-key is "free forever" (LLM Gateway), while buying credits through the gateway carries a 5% fee (pricing). Revenue grows with customers' AI spend.
  • Open source builds trust for infrastructure. Developers can read and self-host the code (LLM Gateway), which lowers the risk of routing all model traffic through a young company.
  • Enterprise upsell. SSO, audit logs, guardrails and an uptime SLA sit in the Enterprise tier (pricing).

What can indie builders learn from LLM Gateway?

  1. Charging a small percentage on usage you already route lets revenue scale with the customer without new sales work.
  2. Open core works for infrastructure: keep the core open and put team and compliance features in a paid tier.
  3. Pick a license on purpose. AGPL plus a commercial directory is one way to stay open while protecting the paid features.

What didn't work for LLM Gateway, and what are the risks?

  • The market is crowded: other self-hosted gateways keep launching on Hacker News, such as Foreman and Mantis in mid-2026.
  • Part of the credit revenue is pass-through spending on model providers, so gross revenue overstates what the company keeps (inference from the 5% fee model on the pricing page).
  • Provider price cuts or provider-native routing could squeeze a thin margin.
  • TrustMRR figures stopped updating on 7 Sep 2026 because the Stripe key expired (TrustMRR).

Sources

Every link cited above, in order of first use.

  1. 1.TrustMRR (trustmrr.com)
  2. 2.LLM Gateway (llmgateway.io)
  3. 3.LICENSE (github.com)
  4. 4.GitHub (github.com)
  5. 5.pricing (llmgateway.io)
  6. 6.Foreman (news.ycombinator.com)
  7. 7.Mantis (news.ycombinator.com)

Related case studies

More in Developer tools →