> PLATFORM_READY

Intelligence
QUANTIZED.

Access Robi's hosted models like lexa-mml and lexa-x1 through a single AI API surface. Top up credits, issue scoped keys, and power everything from agents to analytics without changing your application code.

bash — robi-cli11ms latency
curl -X POST https://api.robiai.com/v1/chat/completions \
-H "Authorization: Bearer robi_sk_..." \
-H "Content-Type: application/json" \
-d '{ "model": "lexa-mml" }'

> Dispatching to EU-west…
{
"id": "chatcmpl-lexa-8x9...",
"usage": { "prompt": 44, "completion": 512 },
"status": "completed"
}

Robi model lineup

Choose between lexa-mml, lexa-x1 and lime — a compact set of hosted models tuned for real workloads.

Scoped keys

Mint project-bound API keys instantly. Apply spending caps, IP allow-lists and workspace-level RBAC.

Token-native billing

Track every prompt and completion token that passes through the platform. Export cost lines per customer or feature.

How Robi AI API fits your stack

Keep your product code simple while Robi handles scale, safety and observability behind the scenes.

Designed for teams running real workloads in production: assistants, agents, copilots and internal tools sharing the same AI surface.

01 · Create

Pick your Robi model

Start with a default Robi model for chat, tools or coding, or define your own named presets with temperature and limits.

02 · Define

Name your models

Create stable model aliases like robi-chat or support-assistant that your applications can rely on release after release.

03 · Ship

Call one API

Point every app to api.robiai.com. Swap models, tweak safety and adjust limits without redeploying.

SDKs & integrations

First-class tooling for the languages and frameworks your team already uses.

TypeScript / NodePythoncURL / HTTPOpenAI-compatible clients

Node

import Robi from "@robi-ai/sdk";

const client = new Robi({ apiKey: process.env.ROBI_API_KEY });

const result = await client.chat.completions.create({
  model: "robi-chat",
  messages: [{ role: "user", content: "Hello from Node" }],
});

Python

from robi_ai import Robi

client = Robi(api_key=os.environ["ROBI_API_KEY"])

resp = client.chat.completions.create(
    model="robi-chat",
    messages=[{"role": "user", "content": "Hello from Python"}],
)

Fuel the grid

Keep your LLM features online with granular credit controls. Auto-recharge when balances dip, so you never drop a production request.

  • > Stripe / credit card billing
  • > Usage dashboards per workspace
  • > Exportable cost & latency traces
CURRENT BALANCE$840.20
EST. RUNTIME~260h @ standard tier
AUTO-RECHARGE THRESHOLD$120.00

Hosted models

A compact lineup of models designed for real workloads, not benchmarks.

Pick a default tier to start, then move individual workloads up or down the stack as you learn more about quality vs. cost.

lexa-x1

Fast & efficient

Great default for everyday workloads: support replies, light reasoning, UI-facing suggestions.

  • > Up to 32K context
  • > Optimized for low latency
  • > Friendly token pricing

lexa-mml

Most capable

Our flagship model for complex reasoning, multi-step tools and higher stakes responses.

  • > Up to 128K context
  • > Strong reasoning and planning
  • > Best for critical paths

lime

Multimodal context

Ideal for long documents and mixed inputs where you care about broader context, not just a single prompt.

  • > Large context window
  • > Handles text plus additional signals
  • > Great for retrieval and analytics

Built for production teams

Robi is opinionated around uptime, safety and predictable costs.

Reliability

Regional redundancy, graceful degradation and clear SLOs.

Observability

Per-request traces with latency, tokens and model metadata.

Controls

Rate limits, soft caps and workspace-level guardrails.

Compliance

Isolation options and data retention policies for regulated teams.