Observed Signal · Feb 18, 2026 · Product Launch · Source: techcrunch · Impact: 3/5 · Sentiment: Neutral
Indian AI Lab Sarvam Launches Game-Changing Open-Source Models
Indian AI lab Sarvam unveiled a new generation of models at the India AI Impact Summit, including 30-billion- and 105-billion-parameter language models, a text-to-speech model, a speech-to-text model and a vision model for document parsing. The models use a mixture-of-experts architecture to reduce compute by activating only a subset of parameters; the 30B model supports a 32,000-token context window for real-time conversational use and the 105B model supports a 128,000-token window for complex reasoning. Sarvam says both models were trained from scratch (the 30B on ~16 trillion tokens) using resources from the IndiaAI Mission with infrastructure support from Yotta and technical support from Nvidia. The company plans to open source the 30B and 105B models and outlined products including Sarvam for Work and a conversational agent platform called Samvaad. Founded in 2023, Sarvam has raised over $40 million from investors including Lightspeed, Khosla and Peak XV Partners.
A notable technical product launch for foundation models with government-backed training and open-source intent; relevant to AI infrastructure and conversational interfaces but not a major platform policy or earnings event.
Track Peak XV Partners Signals & Market Shifts in Real-Time
Polaris7 autonomous intelligence agents track regulatory filings, primary sources, executive changes, and deal flow 24/7. Create your free Explorer workspace to monitor these entities.
Key Takeaways & Evidence Grounding
- Sarvam announced new models including 30-billion- and 105-billion-parameter LLMs, plus TTS, STT and a vision document-parsing model.
- The 30B and 105B models use a mixture-of-experts architecture; 30B supports a 32,000-token context window and 105B a 128,000-token window.
- Sarvam says the 30B model was pre-trained on about 16 trillion tokens; the 105B was trained on trillions of tokens across multiple Indian languages.
- Training resources were provided under India’s government-backed IndiaAI Mission with infrastructure support from Yotta and technical support from Nvidia.
- Sarvam plans to open source the 30B and 105B models and is developing products including Sarvam for Work and the Samvaad conversational agent platform; the company has raised over $40M from Lightspeed Venture Partners, Khosla Ventures and Peak XV Partners.
Connected Companies & Entities
7 Entities mappedOntology Mapping & Concepts
Related Market Signals & Shifts
Recent verified developments and strategic activity across this market segment.
India's Sarvam Unveils Indus AI Chat App Amidst Competition
Sarvam, an Indian AI startup, launched the Indus chat app for web and mobile as the user-facing interface for its newly announced Sarvam 105B large language model. The app, currently in beta on iOS, Android and web and limited to India, supports typed or spoken queries with text and audio responses and offers sign-in via phone number, Google, Microsoft or Apple accounts. Sarvam unveiled both 105B and 30B models at the India AI Impact Summit and announced enterprise initiatives, hardware plans and partnerships with HMD (for Nokia feature phones) and Bosch (automotive). The company has raised $41 million to date from investors including Lightspeed Venture Partners, Peak XV Partners and Khosla Ventures. Sarvam warned of initial compute limits and a possible waitlist; the app also has current UX limitations such as no granular chat-deletion and an always-on reasoning feature.
Sarvam Raises $234M, Becomes India AI Unicorn
Sarvam, a Bengaluru-based AI startup, raised $234 million at a $1.5 billion valuation in a funding round led by HCLTech, the company announced on June 15, 2026. HCLTech is committing $150 million as the lead strategic investor; Bessemer Venture Partners, Khosla Ventures and Peak XV Partners also participated. Sarvam — which earlier in 2026 released open-source 30-billion- and 105-billion-parameter models — is targeting a $300 million Series B raise and plans to invest in next-generation models focused on agentic, coding and cybersecurity use cases. The startup reports large-scale usage across sectors (banking, insurance, government, defense) and says its platforms handle millions of daily interactions and API calls. Founders Vivek Raghavan and Pratyush Kumar aim to commercialize Sarvam’s models together with HCLTech’s enterprise relationships and engineering capacity.
India's Sarvam Unleashes AI for Feature Phones and Beyond
Indian AI startup Sarvam announced plans to deploy its recently released, compact edge AI models to low-end devices including Nokia feature phones, cars and its own smart glasses. At the India AI Impact Summit, Sarvam said its models occupy only megabytes, can run on existing phone processors and can operate offline. The company is partnering with HMD to add a conversational assistant to Nokia/HMD phones, has tuned models with Qualcomm for its chipsets, and is working with Bosch on in-car AI assistants. Sarvam also unveiled Sarvam Kaze, AI smart glasses designed and manufactured in India, which co-founder Pratyush Kumar said will be available in May. The company, previously focused on enterprise voice AI, appears shifting toward consumer use cases.
Track Real-Time Market Signals & Shifts
Set up custom watchlists to receive automated, evidence-grounded executive digests whenever material signals or shifts occur across your tracked landscape.