Baseten
NativePrivateUpdated November 3, 2026AI model-inference/deployment infrastructure
- Website
- www.baseten.co
- CEO
- Tuhin Srivastava
- Founded
- 2019
- HQ
- San Francisco, United States
- Total Funding
- $2.1B
- Employees
- 201-500
Products & Offerings
Baseten Inference Platform (Dedicated Deployments)
Runs custom, fine-tuned and open-source models on dedicated, autoscaling GPUs using the Baseten Inference Stack, deployable in Baseten's cloud, the customer's cloud, or both.
Baseten Model APIs
Production-grade, OpenAI-compatible APIs for frontier open models such as DeepSeek, Llama and Qwen, billed per token.
Baseten Training
Infrastructure and tooling for training and post-training models on customer data before serving them on Baseten.
Truss
Open-source framework for packaging models into deployable, auto-scaling HTTPS endpoints on Baseten.
Agent sandboxes (via Blaxel acquisition)
Secure, persistent execution environments for AI agents' code execution, tool use and sessions, being folded into Baseten's inference platform.
Executive Summary
What they do
- Runs production inference for open-source, custom and fine-tuned AI models at scale
- Offers dedicated GPU deployments and OpenAI-compatible Model APIs for open models
- Expanding into training, post-training and secure sandboxes for AI agents
Why it matters
- Raised a $1.5B Series F in June 2026 at a valuation of up to $13B
- Powers inference for Cursor, Notion, Abridge, Clay, Decagon and other leading AI apps
- Nvidia invested $150M; strategic agreements with AWS and Google Cloud broaden capacity
Best fit
- AI product teams serving latency-sensitive LLM, speech or image workloads in production
- Enterprises moving from closed-model APIs to their own open or fine-tuned models
- Regulated buyers needing self-hosted, hybrid or data-residency-controlled inference deployments
Watch-outs
- Hypergrowth and recent acquisitions may strain support, reliability and roadmap focus
- Faces pricing pressure from Together AI, Fireworks AI and hyperscaler inference services
- Training and agent sandbox offerings are newer and less proven than inference
Latest News
Dealroom · 1 day ago
PositiveKuCoin · 3 days ago
PositiveShiftDelete.Net · 3 days ago
PositiveTekedia · 3 days ago
PositiveBriefs Finance · 3 days ago
PositiveUnlock Full Intelligence
Pro members get the complete briefing
- Complete analyst verdict
- Risk assessment
- All 6 Bluescore dimension scores
- Traction signals
- Bluescore history & momentum tracking
- Unlimited company access
Baseten is one of the leading independent inference platforms. It has heavy funding, Nvidia backing and production customers such as Cursor, Notion and Abridge, so any team running open or custom models in production should consider it now. Benchmark latency, cost per token and SLAs against Together AI, Fireworks AI and your hyperscaler's native option, and get capacity and support commitments in writing given how fast Baseten is growing.
Signal Provenance
Latest observation date and source for each signal category
| Signal Category | Last Updated | Source |
|---|---|---|
Funding Signals Last funding round, Funding stage | Nov 2, 2026 | Montclare research agents |
Hiring Velocity Headcount, Headcount growth | Nov 3, 2026 | Montclare research agents |
Technical Output Research papers, GitHub activity | Nov 3, 2026 | Montclare research agents |
News Sentiment News mentions, News sentiment | Nov 3, 2026 | Montclare research agents |
Partnership Signals Enterprise partnerships, Distribution partners | Nov 3, 2026 | Montclare research agents |
Revenue Indicators ARR estimate, Revenue growth | Nov 3, 2026 | Montclare research agents |
Leadership Signals CEO tenure, Founder-led | Nov 3, 2026 | Montclare research agents |
Competitive Position Category rank, Market share | Nov 3, 2026 | Montclare research agents |
Why this Bluescore?
6 dimensions, each scored 0–10 against all — companies. Each adds its weight × score to the Bluescore — see methodology