Aragt
Contact
Platform

Everything you need to build with AI, in one platform.

Foundation models, agents, hosting, and developer tools,designed for teams building serious AI systems for the real world.

Get started

Ship your first request in minutes.

A single SDK, a single API key. Works with our models and popular open-source models.

Request accessBrowse open models
quickstart.ts
import { Aragt } from "@aragt/sdk";

const client = new Aragt({ apiKey: process.env.ARAGT_API_KEY });

const response = await client.messages.create({
  model: "aragt-core-1",
  messages: [
    { role: "user", content: "Draft a market summary in Swahili." }
  ],
  tools: [{ name: "search", type: "web_search" }],
});

console.log(response.output_text);
Capabilities

The building blocks.

Everything is available as a managed service, an API, and,where it makes sense,as an open-source release.

  • API

    AI APIs

    A single, versioned interface to our language, multimodal, embedding, and speech models.

    • ›Streaming chat and completions
    • ›Structured outputs
    • ›Function calling & tools
    • ›Fine-grained rate control
  • Models

    Foundation Models

    General-purpose and specialized models trained by our research teams.

    • ›Aragt Core,general reasoning
    • ›Aragt Code,coding & agents
    • ›Aragt Multimodal,vision + text
    • ›Aragt Speak,ASR & TTS
  • Agents

    Agent Platform

    Build, evaluate, and deploy long-running agents with tool use, memory, and oversight.

    • ›Tool servers & sandboxing
    • ›Memory & workspace primitives
    • ›Human-in-the-loop review
    • ›Evaluation harness
  • Hosting

    Model Hosting

    Serve our models, fine-tuned variants, or open-source models on managed infrastructure.

    • ›Autoscaling inference
    • ›Batch endpoints
    • ›Private deployments
    • ›Regional data residency
  • Cloud

    Cloud AI Services

    Data, training, and evaluation pipelines that turn your workflows into production AI.

    • ›Managed fine-tuning
    • ›Vector & retrieval services
    • ›Data curation tooling
    • ›Observability & logging
  • Tools

    Developer Tools

    SDKs, CLIs, and IDE integrations that make building on Aragt fast and pleasant.

    • ›Typed SDKs (TS, Python, Go)
    • ›CLI + local runner
    • ›Prompt & eval studio
    • ›Playgrounds
Evaluation

Built-in evaluation for the systems that ship.

Every capability in the platform is paired with tooling to measure it,model quality, agent success rates, latency, cost, and safety metrics.

Teams can define their own evaluations, wire them into CI, and gate deployments on the metrics that matter for their use case.

  • Golden datasets
  • LLM-as-judge
  • Rubric-based scoring
  • Trace inspection
  • A/B routing
  • Regression alerts