India’s homegrown artificial intelligence ecosystem is making a strong push on pricing, with BharatGen and Sarvam AI launching foundation models that cost only a fraction of comparable offerings from global AI companies. According to pricing reviewed by industry reports, BharatGen’s Param-2 is priced at ₹5 per million output tokens, while Sarvam AI charges ₹10 for its 30-billion-parameter model and ₹16 for its flagship 105-billion-parameter model. By comparison, OpenAI’s GPT-5 Mini costs around ₹191 per million output tokens, while Google’s Gemini 3.5 Flash is priced at approximately ₹858.
The aggressive pricing strategy is aimed at accelerating AI adoption across India, particularly for enterprises building multilingual applications and services. Analysts say the low prices are made possible in part by government-backed GPU subsidies under the IndiaAI Mission, though questions remain about whether such pricing can be sustained over the long term without continued policy support.
BharatGen and Sarvam Undercut Global AI Rivals
The pricing gap between Indian and global AI providers is significant.
| Model | Output Token Price (per million) |
|---|---|
| BharatGen Param-2 | ₹5 |
| Sarvam 30B | ₹10 |
| Sarvam 105B | ₹16 |
| OpenAI GPT-5 Mini | ₹191 |
| Google Gemini 3.5 Flash | ₹858 |
The figures illustrate how Indian AI developers are prioritizing affordability to drive domestic adoption.
Why Indian Models Are So Cheap
Several factors contribute to the lower pricing.
Key reasons include:
- Government subsidies for AI computing infrastructure.
- IndiaAI Mission support for GPU access.
- Focus on Indian-language use cases rather than broad frontier AI capabilities.
- Lower operating costs for targeted deployments.
- Open-source and public-interest development models.
Industry experts note that subsidized compute has helped reduce inference costs for domestic AI developers.
Focused on Indian Languages
Unlike many frontier AI models designed for global use, BharatGen and Sarvam have concentrated on India’s multilingual ecosystem.
Key strengths include:
- Support for multiple Indian languages.
- Code-mixed language understanding.
- Government and enterprise applications.
- Translation and speech capabilities.
- Regional AI deployment.
This specialization allows the companies to optimize models for local use cases rather than competing directly across every AI benchmark.
Price vs Capability
| Area | Indian Models | Global Frontier Models |
|---|---|---|
| Cost | Much lower | Higher |
| Indic language support | Strong | Broad but less specialized |
| Frontier reasoning | Improving | Generally more advanced |
| Global enterprise adoption | Emerging | Mature |
Analysts caution that while Indian models are dramatically cheaper, they are not yet considered direct replacements for the most capable frontier models from OpenAI, Google, or Anthropic across all tasks.
Can the Pricing Last?
The biggest question is long-term sustainability.
Challenges include:
- Dependence on subsidized GPU infrastructure.
- Rising inference costs as usage grows.
- Ongoing investment required for model training.
- Competition from global AI providers.
- Need for continuous model improvements.
If government support is reduced, maintaining today’s pricing levels could become more difficult.
Why This Matters
Affordable inference pricing could significantly accelerate AI adoption across India.
Potential beneficiaries include:
- Startups building AI applications.
- Government digital services.
- Educational institutions.
- Regional language platforms.
- Small and medium-sized businesses.
Lower API costs reduce one of the biggest barriers to deploying generative AI at scale.
Outlook
The launch of ultra-low-cost foundation models by BharatGen and Sarvam AI marks an important milestone in India’s sovereign AI ambitions. By pricing inference dramatically below global competitors, the two organizations are making advanced AI more accessible to developers and enterprises, particularly those building applications for India’s multilingual market.
However, pricing alone is unlikely to determine long-term success. Industry observers note that global AI leaders still maintain an advantage in overall model capability, research scale, and ecosystem maturity. The next challenge for India’s AI ecosystem will be combining competitive performance with affordable pricing while reducing reliance on government-supported infrastructure.
What It Means for India’s AI Industry
The pricing strategy demonstrates India’s growing focus on building a domestic AI ecosystem tailored to local needs rather than simply competing head-on with global frontier models. Affordable APIs could encourage faster AI adoption among startups, enterprises, and public-sector organizations that were previously constrained by inference costs.
If BharatGen and Sarvam continue improving model quality while maintaining competitive pricing, they could help establish India as a significant player in multilingual AI infrastructure. At the same time, long-term competitiveness will depend on sustained investment in research, computing infrastructure, and model innovation—not just low prices.
Get the day’s top stories in your inbox
One concise email. No spam, unsubscribe anytime.
