Chat, vision and listening
Chat completions with text, image and audio input, streaming and tool calls, plus transcription of recordings.
LeanModel
Text, image, voice and video intelligence, one key away.
LeanModel gives your agents one way to think, see, speak and show. Seven model catalogues sit behind one OpenAI-compatible API: chat with text, images and audio, image generation and editing, speech, music and video generation, talking avatars over WebRTC and realtime voice over WebSocket, with model tiers matched to the job and every call on the same bill as everything else.
01
Drop-in compatible
Point any OpenAI-compatible SDK at LeanModel and keep the code you have. Avatars and realtime voice use the same key over WebRTC and WebSocket.
02
The right model for the job
Every catalogue has standard tiers with a published price, and operators can add more. Pick a tier for cost or for quality.
03
One bill, not seven
No separate provider accounts. Every token, image, character, second and minute draws from your LeanTools credits.
Capabilities
What your agents can do with LeanModel.
Image generation and editing
Create images from a prompt, or edit the images you send, in a single call.
Speech synthesis
Give your agents a voice with natural text-to-speech in the voice you describe.
Music generation
Turn a prompt and optional lyrics into an audio clip.
Video generation
Render short video clips from a prompt, at the length, resolution and orientation you choose.
Talking avatars
Stream a live avatar over WebRTC and have it speak whatever your agent says.
Realtime voice
Speech-to-speech conversations over one WebSocket, with tool calls and transcripts as they happen.
Model tiers
Match each task to a tier, from fast and economical to most capable. Every tier lists its capabilities, operations and prices.
Spend controls
Per-key monthly caps and per-tool budgets keep every agent within limits.
How your agent uses it
Keep your SDK. Change one line.
Point any OpenAI-compatible client at LeanTools and your agents think, see and speak on the same key and the same bill as every other product. Avatars and realtime voice take the same key over WebRTC and WebSocket.
API referencefrom openai import OpenAI client = OpenAI(base_url="/api/v1", api_key="lt-…") reply = client.chat.completions.create( model="leanmodel-flash", messages=[{"role": "user", "content": "Summarise today's incidents."}], ) print(reply.choices[0].message.content)
Use cases
Where teams put LeanModel to work.
Summaries and drafting
Condense long threads, draft replies and rewrite in any tone.
Voice interfaces
Hold realtime voice conversations, transcribe calls and answer back with natural speech.
Media on demand
Generate images, music, video and live avatar presenters for products, campaigns and prototypes.
Which SDKs can we use?
Any OpenAI-compatible SDK or HTTP client. Point the base URL at LeanTools, use your LeanTools key, and keep the rest of your code. Avatar streams and realtime voice use plain REST and WebSocket calls with the same key.
Which models are available?
Models are offered in seven catalogues (text, image, speech, music, video, avatar and realtime audio), each with standard tiers at a published price. Your console lists the tiers your workspace can use, and the API lists them with their capabilities and rates.
Do you keep prompts or outputs?
No. LeanTools stores only the measurements needed for billing, never request or response content, and generated media is deleted once it has been returned to you.
How is it priced?
By tier: per million tokens for text and realtime audio, per image, per 1,000 characters of speech, per second of music or video, and per connected minute of an avatar stream.
How long do music and video calls take?
The response waits until the clip is rendered, which can take a few minutes. Allow a request timeout of at least five minutes in your client.
Better together
Pairs well with LeanModel.
LeanMemory
A memory that learns from use: structured, unstructured and event data in one place, with recall that gets more accurate the more your agents work.
LeanVision
Continuous vision monitoring for CCTV. Describe what matters and get events the moment it happens.
LeanConnect
Connect agents to the apps your business runs on, with credentials kept safe and every write under your control.
Start building with LeanModel.
One endpoint, one key and every LeanTools product, from your first call to production.