Intelligence Suite
EN

Products

Five Products.One governed platform.

From compute to agents, everything runs through the same connection, the same policy engine and the same bill.

AI Compute

Maximum tokens per second from every GPU.

Reserved GPU clusters, on-demand GPU, AI storage and networking, run on Indosat's data centres, power and national network.

Reserve Capacity

  • Reserved GPU clusters

    Dedicated capacity for long-running training and inference.

  • On-demand GPU

    Burst capacity when workloads spike.

  • AI storage & networking

    High-speed interconnect and storage built for AI.

  • Powered by ZanKore

    AI-factory infrastructure on NVIDIA

AI Token Marketplace

Every model through one governed API.

One marketplace to browse, compare and access every model. Auth, rate limits, token metering, usage dashboards and spend controls are built in.

Get endpoint accessBrowse the catalogue
Time to first token at P95
< 3s
Availability
99.9%
Throughput
Guaranteed
Models, launch to roadmap
100+ → 300+

  • Compare and choose

    See models side by side and pick the right one for each task.

  • Compare on price

    Weigh cost per task across comparable models before you commit.

  • Switch without rebuilding

    Change models through the same API, with no re-integration.

  • Multimodal by default
  • Serverless inference
  • OpenAI-compatible

AI Factory Platform

Live Nov 2026

Build, adapt and serve AI as a managed service.

Bring models and agents, ground them in your data, and deploy privately, without owning the serving stack. Your data is your differentiation; the Suite lets you use it without exposing it.

Join the early-access list

  • MaaS

    Model-as-a-Service

    Bring any model and serve it privately.

  • AaaS

    Agent-as-a-Service

    Governed agents with tools, memory and workflows, ready to run.

  • RAG

    Retrieval

    Ground models in your own documents, so answers cite real, current sources.

  • Guard

    Guardrails

    Prompt controls, output validation and PII redaction on every request.

  • Tune

    Fine-tuning

    Adapt models to your data, under the same governance and residency.

AI Industry Solutions

Ready-to-use AI, built for your industry.

Start from a proven use case and work with us to tailor it to your needs.

Contact Us

Sahabat-AI

AI that speaks Indonesia’s languages.

Sahabat-AI is an open-source large language model family built for Bahasa Indonesia and local languages. It runs on Indosat infrastructure in Indonesia and is available through the AI Token Marketplace.

Try Sahabat AI

  • Open-source and free

    Model weights are free to download and use, so you can inspect, fine-tune or self-host.

  • Built for Bahasa Indonesia

    Trained for Bahasa Indonesia and local languages: Javanese, Sundanese, Balinese and Batak.

  • Models for every size

    From 8B and 9B models to a 70B flagship, built on Llama and Gemma.

  • Hosted in Indonesia

    Runs on local infrastructure, so data and processing stay in-country.

Find the right entry point for your workload.

Book a demoSee pricing