Integrations · Agent frameworks
Build production agents on Vidman AI.
LangChain, LlamaIndex, CrewAI, PydanticAI, Strands Agents — every orchestration framework on that list takes an OpenAI-compatible endpoint. Aim one here and your agents run on the same API, serverless or on your own dedicated deployment.
from openai import OpenAI
client = OpenAI(
base_url="https://api.vidman.ai/v1",
api_key="sk-vidman-...",
)
resp = client.chat.completions.create(
model="deepseek-v4-flash", # any served model id
messages=[{"role": "user", "content": "Summarize my inbox"}],
)Three values — base URL, API key, model id. Every framework below takes the same three, just in its own config.
Frameworks
Where the three values go
Each framework keeps its own config format — the values never change. Exact file paths vary between versions; the pattern below is the one every version accepts.
LangChain
Build LLM applications with powerful orchestration and tool integration.
ChatOpenAI — set base_url and api_key on the model, pick any served model id.
LlamaIndex
Efficient data retrieval and document indexing for LLM-based agents.
OpenAI LLM — set api_base and api_key, then index and query with any served model.
CrewAI
Orchestrate collaborative multi-agent systems for complex tasks.
LLM config — base_url and api_key per agent, or once for the whole crew.
PydanticAI
Type-safe AI agent development with Pydantic validation.
OpenAIModel — pass base_url and api_key to the model constructor.
Strands Agents
Modern agent orchestration with seamless OpenAI-compatible integration.
Model provider config — an OpenAI-compatible endpoint with your key and model id.
AWS AgentCore
A managed runtime for deploying production AI agents, with bring-your-own model providers.
Model provider settings in the AgentCore console — OpenAI-compatible endpoints supported.
Not listed? If it accepts an OpenAI-compatible endpoint, the same three values apply. Per-framework walkthroughs are being verified and will be linked here as they pass.
Why Vidman AI
The framework builds the agent. Vidman AI serves the model.
The framework keeps its loop
Memory, tools and orchestration remain in the framework you already picked. Vidman AI provides the model behind one OpenAI-compatible endpoint — everything else stays put.
Agents burn tokens. Pay by the token.
An always-on agent is a high-volume workload, and serverless inference meteres per million tokens at the catalog rates — no seat licences, no minimums.
Your weights, the same three lines.
Train a model on dedicated GPUs and serve it on a dedicated endpoint behind the same API. Nothing in the framework config moves — only the model id does.
Questions teams ask before switching
Which orchestration frameworks run on Vidman AI?+
Anything that accepts an OpenAI-compatible endpoint: LangChain, LlamaIndex, CrewAI, PydanticAI, Strands Agents, AWS AgentCore, plus the OpenAI SDKs themselves. The integration surface is those three values — base URL, API key, model id.
Does my framework code have to change?+
No. You change where the framework looks, never what it does: set the base URL to the Vidman AI endpoint, drop in your key, pick a served model id. The framework keeps its own loop, tools and memory.
Can the framework use a model I fine-tuned here?+
Yes. A finished checkpoint can sit on a dedicated endpoint behind the same OpenAI-compatible API. The framework config is identical — only the model id moves, and the weights behind it are yours.
How does framework traffic get billed?+
By the million tokens, at the same input and output rates the catalog publishes. No per-seat or per-agent fees exist, and cached input is cheaper wherever the model offers it.
Aim your framework at Vidman AI tonight.
Get 50% extra on your first wallet top-up.
Per-million-token billing · Zero data retention · Your weights stay yours