The archive · Developer & Business Tools · Product decision · 2023–2026
BerriAI bets every AI app needs a gateway; LiteLLM passes 51k stars
BerriAI extracted LiteLLM from its own integration pain in 2023; the open-source AI gateway passed 51k GitHub stars and 1B+ proxied requests by 2026.
LiteLLM (BerriAI)
What the business is
LiteLLM is an open-source AI gateway from BerriAI: a Python SDK and self-hostable proxy server that lets teams call 100+ LLM providers through one OpenAI-compatible interface, with cost tracking, guardrails, load balancing and logging.
Starting capital：$1.6M seed round in 2023, co-led by Y Combinator with participation from Gravity Fund and Pioneer Fund.
How it started
BerriAI was founded in 2023 by Ishaan Jaffer and Krrish Dholakia and accepted into Y Combinator's Winter 2023 batch. On 2023-08-24 YC publicly launched LiteLLM: an open-source package to call 100+ LLM APIs using the OpenAI format, with standardized inputs, outputs and exceptions, plus logging integrations. The team raised a $1.6M seed co-led by Y Combinator with Gravity Fund and Pioneer Fund.
What happened
LiteLLM grew from a library into a category: the self-hostable Proxy Server (AI Gateway) added virtual keys, per-key and per-team budgets, spend tracking, routing with fallbacks and cooldowns, guardrails, an admin dashboard and, later, an MCP gateway for agentic tool access. Adoption compounded through open source — InfoWorld reported 20,000+ stars by mid-2025, and by mid-2026 the repo had roughly 51,500 stars, 1,372 tagged releases, 1B+ proxied requests and 240M+ Docker pulls, with production users including Netflix, Stripe, Lemonade, Rocket Money and Adobe. Founderland reported 50k+ stars and $7M ARR in August 2026, after the company navigated security challenges earlier that year.
How it ended up
Still scaling: the repository remained live with frequent releases (v1.89.4, 2026-06-25), the gateway added agent (A2A) and MCP support, and the company offers LiteLLM Cloud as the commercial layer.
Background
LiteLLM is the open-source AI gateway built by BerriAI, founded in 2023 by Ishaan Jaffer and Krrish Dholakia and accepted into Y Combinator's Winter 2023 batch. The product was extracted from the founders' own pain: while building a chat-with-your-data SaaS, they found supporting multiple LLM backends — each with different SDKs, auth patterns, request formats and error models — a substantial engineering burden.
Their bet was that every AI application would end up calling many providers, so a unified interface would become standard infrastructure. Launched through YC on 2023-08-24, LiteLLM let developers call 100+ LLM APIs using the OpenAI format, making a provider switch a one-line change of the model string rather than a client rewrite.
Adoption compounded: InfoWorld reported 20,000+ GitHub stars by mid-2025; by mid-2026 the repo had roughly 51,500 stars, 9,200 forks and 1,372 tagged releases, with more than one billion requests proxied and 240M+ Docker pulls. Production users named include Netflix, Stripe, Lemonade, Rocket Money and Adobe. The gateway added virtual keys, budgets, spend tracking, load balancing with fallbacks, guardrails, an admin dashboard and later MCP and A2A agent support.
The company raised a $1.6M seed co-led by Y Combinator with Gravity Fund and Pioneer Fund in 2023, and Founderland reported 50k+ stars and $7M ARR in August 2026. As of September 2026 the project is still scaling, with LiteLLM Cloud as the commercial layer and the open-source repository releasing nearly daily.
What has to be true
- Multi-provider integration was a universal, repeated pain — every team using more than one LLM rebuilt the same translation, error-handling and streaming plumbing.
- Standardizing on the OpenAI format made adoption frictionless: existing SDKs and client code kept working with a base-URL or model-string change.
- Open-sourcing the core turned users into contributors and distribution, while enterprise controls (virtual keys, budgets, guardrails) created the paid tier.
- The gateway sits at the chokepoint of AI spend and traffic, so usage data, routing and cost controls compound into a defensible infrastructure position.
What can be applied
A boring-but-essential layer can compound into infrastructure: standardize on a format developers know, keep the core open, and monetize the controls — keys, budgets, routing.
Aftermath
As of September 2026, LiteLLM is still scaling: the GitHub repository (BerriAI/litellm) is live with frequent releases (v1.89.4 on 2026-06-25 per AI Wiki), ~51,500 stars, 1B+ proxied requests and 240M+ Docker pulls, and production users including Netflix, Stripe, Lemonade, Rocket Money and Adobe. Founderland reported 50k+ stars and $7M ARR in August 2026, noting the company navigated security challenges earlier that year. The commercial layer is LiteLLM Cloud, alongside self-hosted deployment via Docker, Helm and Terraform.
Sources
- LiteLLM | AI Wiki
- LiteLLM Raises $1.6M Seed to Scale AI Gateway Platform
- BerriAI/litellm — The fastest, litest AI Gateway
spotted an error? The archive wants to know.
Your turn
You just read one. Describe what you are building, and see who is betting on the same thing.
Free account · 3 free questions · no card