Empowering Developers through Open-Source AI
summary
Samanova runs open-source models on proprietary reconfigurable data units that use a three-tier memory and 16‑chip nodes to cut inference latency; a single node hosts 10–20 fine‑tuned models per node and sustains world‑record tokens‑per‑second throughput for agentic workflows. A customer (MI Tai) converts many enterprise checkpoints into low‑latency voice and inference services, reporting dramatic latency drops; claimed cost is roughly $0.22 versus $3 per million tokens against a closed‑API alternative.
AI-written summary · about
chapters (10)
- 0:00Introduction to Sam and AI Engineering
- 0:26Understanding the AI Landscape
- 0:59Open Source vs Closed Source Models
- 1:54Advantages of Open Source Models
- 3:31Evolution of AI Capabilities
- 5:06Importance of Speed in AI Systems
- 6:07Samano Cloud and Hardware Innovations
- 8:23Agent Tech Applications and Integrations
- 10:56Real-World Use Case: MI Tai
- 12:47Conclusion and Q&A
related talks
35mUsing coding agents with open models
17mOpen Innovation: Unlocking AI's Potential
30mGenerative AI in action: from POC to business value
19mDeveloper Centric AI
29mOpen Source Sovereign AI: Building Enterprise GenAI Platforms for Control & Comp...
10mThe Future for AI-First Enterprises is “Open” - How to Succeed and Not...
15mThe Open Ecosystem for the Agentic Age
22mSponsored Demo: Cut the Bill, Not the Brains. AI Coding Agents That Ship More and...