One API for every AI model.
Call Claude, GPT, Gemini, DeepSeek and more through a single OpenAI- and Anthropic-compatible endpoint — billed per token, with automatic fallbacks and your own provider keys. Then go further with Chat and Memory: value-adds a plain relay can't offer.
ⓘ This service is for learning and technical research only. Please confirm compliance in your region before using; see thefor details.
from openai import OpenAI
client = OpenAI(
base_url="https://www.getllm.ai/v1",
api_key="gl-live-...", # one key, every model
)
resp = client.chat.completions.create(
model="claude-sonnet-4-6", # swap to gpt-4o, gemini-… freely
messages=[{"role": "user", "content": "Hello!"}],
)Gateway
CorePoint any OpenAI- or Anthropic-compatible SDK at GetLLM and keep your code. One key reaches every model; we handle routing, fallbacks, and per-token billing at published rates.
Claude, GPT, Gemini, DeepSeek and more — one account, one API key. Compatible with the OpenAI / Anthropic SDKs you already use.
Automatic context caching, real-usage billing, and routing that picks the best model — so daily AI calls stop being a budget headache.
Every call is logged and reconcilable. Pay-as-you-go, billed per token; your balance never expires.
https://www.getllm.ai/v1https://www.getllm.aiThe list below is what the gateway can serve today — usable directly from your API key. Fetched live from the backend; no login required.
Loading…
Memory
· In developmentA memory system that sits on top of every large model. The things you've said, the documents you've uploaded, the questions you've asked — none of it is lost when the underlying model changes. Beyond a conversation workspace, Memory offers APIs so third-party apps can embed this memory layer into their own products.
Send messages, files, and facts to the Memory API — Memory extracts what matters.
It builds a structured, per-user memory graph that survives across models and sessions.
Every new call is enriched with the right context — cited, scoped to the user, yours to export.
What you said to Claude today carries over to GPT tomorrow and to a local model the day after. Your context belongs to you, not to a single model.
Upload notes, papers, contracts, archives — Memory organizes them into a retrievable knowledge layer with precise citations, so the AI has actually "read your files."
A GraphRAG-backed domain knowledge graph that weaves authoritative materials and their relations into traceable answer chains, each conclusion citing its source. The first verticals will be announced on launch day.
More than a relay
Value-addMost gateways just forward your request. GetLLM adds a product layer on top — so you ship faster, and your users get something commodity resellers can't match.
The no-setup way to use GetLLM: an AI chat that talks to any model we route — and remembers you across conversations. Upload images and video, and generate images and video right in the thread. Just sign in and start typing.
Try ChatStart a new conversation and it still knows your context — powered by the Memory layer, not a fixed prompt window.
Drop in a screenshot or clip and ask about it — images are read by vision models; video is attached to the thread.
Switch to image or video mode and create media from a prompt, saved straight into the conversation.
Get started in three steps
Create an account with Google, GitHub, or email — no card required to start.
Top up once. Credits never expire and work with any model or provider.
Mint a gl-live key and point your SDK's base_url at GetLLM — fully OpenAI- and Anthropic-compatible.
What's new
All changes →Start building in minutes
Create a key, point your SDK at GetLLM, and call any model. Scale into memory and chat whenever you're ready.