India's sovereign AI inference — DPDP compliant, zero retention by default.Explore models
Zero retention by default
Drop-in OpenAI & Anthropic compatibility
Efficient models, lightning-fast
Latest models deployed. Fast
Model Catalogue
10+ frontier models. One OpenAI-compatible API. Billed in INR.
tm/glm-5-2
NewOpen-weight frontier model with strong coding, agentic performance and a 1M context window.
tm/minimax-m3
PopularOpen-weight model built for coding, long-context agent work and multimodal input.
tm/deepseek-v4-pro
FlagshipDeepSeek's most capable model for complex reasoning, coding, and agentic workflows.
tm/kimi-k2-6
FastOptimised for low-latency responses with strong multilingual support.
tm/kimi-k2-7-code
CodeSpecialised code assistant tuned for software-engineering tasks.
tm/deepseek-v4-flash
BudgetMaximum cost-efficiency for high-volume production workloads.
Built for India's Regulatory Landscape
We don't just know compliance — we're engineered around it. Every workload runs within India's sovereign infrastructure.
DPDP Act 2023
Digital Personal Data Protection Act compliant infrastructure with data fiduciary controls
MeitY Guidelines
Aligned with Ministry of Electronics & IT data governance and AI framework policies
RBI Data Localisation
Payment and financial data stays within Indian borders per Reserve Bank of India mandates
SEBI Compliance
Capital markets AI workloads meeting Securities and Exchange Board of India requirements
Certified infrastructure
Runs on AWS's Mumbai (India) region, which holds ISO 27001 and SOC 2 Type II. Tensor Machine does not itself hold these certifications yet.
CERT-In Aligned
Cybersecurity posture aligned with CERT-In reporting requirements and incident response
IP Ownership
Your IP is yours. Not training data.
We never train on your prompts or outputs — and content logging is off unless your org turns it on, enforced at the edge.
Power Every Vertical
Every workload below runs inside India by default. That settled — here's what teams actually build, from BFSI to healthcare, government to startups.
Secure Code Assistance
Code generation, review, and refactoring over large proprietary codebases — in your IDE, your CI, and your internal tools.
Regulated Industry AI
Document processing, underwriting support, and clinical summarisation for BFSI, insurance, and healthcare — the sectors where model choice is usually constrained.
Agentic Systems
Multi-step agents with full observability — tool calls, retries, and audit trails you can actually inspect when a workflow misbehaves.
Enterprise RAG
Retrieval-augmented generation over your internal knowledge base, with Hindi and regional-language document support out of the box.
Conversational AI
Customer-facing AI in 22+ Indian languages. Indic script support. Low-latency responses optimised for Indian network conditions.
Financial Analytics
Market research, filings summarisation, and fraud detection over BSE/NSE and payments data — at the speed a trading day demands.
from openai import OpenAI
client = OpenAI(
api_key="your-tensormachine-api-key",
base_url="https://edge.tensormachine.ai/v1"
)
response = client.chat.completions.create(
model="deepseek/deepseek-v4-pro",
messages=[{"role": "user", "content": "Hello"}],
stream=True
)
for chunk in response:
print(chunk.choices[0].delta.content, end="")SWITCH IN 60 SECONDS
Switch to Tensor Machine
with one line of code
Spin up any of our models on India-sovereign infrastructure in minutes. No DevOps expertise required — just pick your model and go. Change only the base_url.
Compliance shouldn't pick your models
Regulated teams don't reject frontier AI. They reject the data-transfer review attached to it.
Compliance was picking your models
Data-localisation rules disqualify providers that can't serve from inside India. The shortlist shrinks before the evaluation even starts — and the model you wanted is usually the one that gets cut.
Sovereignty, without the shortlist
Every model we route to is served from India-region compute. Residency isn't a clause in a policy document — it's a per-tenant guardrail in the request path, and a provider that can't satisfy it is never offered.
Room to experiment again
Prompts and completions aren’t stored unless you ask for them to be: content logging is off by default and enforced at the edge. No cross-border data-transfer review standing between a team and a prototype.
We're onboarding design partners across BFSI, healthcare, and government. Talk to our team about your workload.
Pay in rupees. File your GST.
AI infrastructure that fits an Indian budget process — not a dollar bill your finance team has to explain every month.
Priced and billed in rupees
Per-token rates are quoted in INR and charged in INR. No dollar card, no forex conversion on your statement, no exchange-rate surprise between quote and bill.
GST invoices, filed as-is
Proper GST tax invoices against your GSTIN — sequentially numbered, immutable once issued, with SAC codes and the CGST/SGST/IGST split your finance team expects.
Prepaid, with hard limits
Top up a wallet and set spend limits per team and per member. Limits are enforced at the edge on every request, so a runaway job stops at the cap — not at the invoice.
FOR REGULATED ENTERPRISES
Sovereign. Compliant.
Mission-critical.
Deploy AI with confidence in regulated Indian industries. DPDP Act compliant. Zero exposure of your sensitive data.
- ISO 27001 / SOC 2 certified infrastructure (AWS)
- Content logging off by default
- Dedicated India compute
- DPDP Act compliant data processing
- DPA and SLA available
- RBI / SEBI / IRDAI aligned
FOR INDIA'S BUILDERS
Drop-in OpenAI compatibility.
Zero friction.
Change one line of code. Access 10+ frontier models through one OpenAI-compatible API, billed in INR.
- Full OpenAI-compatible API
- 10+ open & frontier models
- India data residency included
- Generous free tier to start
- Single API key for all models
Intelligence
View all posts
Tensor Machine Is Live: Sovereign AI Inference, Built for Bharat
After two years of building in stealth, India's first dedicated sovereign AI inference platform is live today. No prompt data leaves Indian borders. No foreign exchange markup. No DPDP grey areas. Here is what we built, and why it matters for Indian enterprises right now.
Read more
₹80 Lakhs vs ₹18 Lakhs: The True Cost of AI Inference for Indian Teams
We modelled the total cost of AI inference for a mid-sized Indian SaaS company processing one billion tokens per month. US-based providers cost ₹80 lakhs all-in once you account for FX exposure, GST reverse charge, and compliance overhead. The same workload on Tensor Machine costs ₹18 lakhs. Here is the full breakdown.
Read more
DPDP Act 2023 and Your AI Product: What Engineering Teams Actually Need to Know
India's Digital Personal Data Protection Act 2023 has direct implications for any product that processes personal data through AI inference. Most engineering teams do not know where their prompts go, who processes them, and whether they qualify as personal data under Indian law. This guide breaks it down in plain terms.
Read more
Stay ahead of India's AI curve
Get updates on DPDP compliance, new model releases, and sovereign AI. No spam, and you can unsubscribe at any time.