Open-weight models.
Coding, agents, vision, reasoning and Indian languages. Live on GPUs in India.
Qwen3.8-27B
Gemma 4 31B IT
GLM-5.3
DeepSeek V4.1 Flash
Nemotron 3.5 Super VL 120B A12B
For research groups
Spend your grant on research, not on model bills.
Open-weight models on GPUs in India, billed in rupees. Research groups can apply for discounted access.
We work with research groups as a collaborator.
| Model | Tokens | Time | Key |
|---|---|---|---|
| Qwen3.8-27B | 1,842 | 14:02:11 | lab-genomics···4f2a |
| Gemma 4 31B IT | 3,107 | 14:02:19 | lab-vision···91c0 |
| GLM-5.3 | 926 | 14:02:27 | lab-genomics···4f2a |
Where we want to help
- AI research
- Age tech
- Longevity
- HealthcareMulti-omics, cancer, eye and geriatric research
One key for your whole lab.
The models your work needs, without rewrites.
No rewrite.
Using the OpenAI SDK? Change the base URL and the key.
client = OpenAI(
base_url = INDICSTACK_BASE_URL,
api_key = INDICSTACK_API_KEY)Medical models.
MedGemma and TxGemma, deployed on request on GPUs in India.
Clinical use needs your own validation. Model capability is not clinical or regulatory approval.
Search your papers.
Embedding and reranking models, on request.
Need another model?
Name it. Dedicated GPU deployments in India are on request.
Qwen3.8-27B
Gemma 4 31B IT
Open-weight modelDedicated · India
Model Scout agent.
Tracks new open models on HuggingFace, arXiv and GitHub.
HuggingFacearXivGitHub
New models
Every call on record.
Model, tokens, time and key, logged for each call.
| Model | Tokens | Time | Key |
|---|---|---|---|
| Qwen3.8-27B | 1,842 | 14:02:11 | lab-genomics···4f2a |
| Gemma 4 31B IT | 3,107 | 14:02:19 | lab-vision···91c0 |
| GLM-5.3 | 926 | 14:02:27 | lab-genomics···4f2a |
Your data stays yours.
Never used for training.
Prompts and completions on India-hosted models.
Not stored.
API prompts, files and completions are deleted after the response.
Metadata only.
Time, model, tokens and key ID, kept for 12 months.
You own it.
What you send and what you get back.
Runs in India.
India-hosted models run on GPUs in Indian data centres.
Your responsibility.
Your institution is the Data Fiduciary under the DPDP Act.
Documents on request.
Sub-processors, data flow and security answers.
Who can apply
- Research groups at a university, institute or hospital
- Led by a named principal investigator
- Working in one of the focus areas
What we look at
Group
PI, institution and official email
Question
Your research question and area
Data and ethics
What you send, and ethics approval for human data
Fit
A live model or an on-request one
Scale
Expected use and dates
Apply in minutes
- 01
Apply
Contact form, research topic selected
- 02
We confirm
- 03
We review
- 04
Terms, if you qualify
- 05
You start
Don't include participant data in your application.
Questions
Is the programme open?
We're opening it now. Apply and we'll follow up.
How much is the discount?
We're setting the terms now. Apply and we'll follow up.
Can we start before a decision?
Yes. The Free Trial gives you 15 days on Qwen3.8-27B. No card needed.
Can we use the medical models with patients?
Only after your own clinical validation.
Where do the models run?
India-hosted models run on GPUs in India. Global models run outside India. Check the label on each route.