baseten.coCrawled Oct 16, 2026
The Baseten Frontier Gateway is described as a path from weights.
Use Chains to orchestrate inference workflows across multiple models, with a framework designed for performance.
Run mission-critical inference at massive scale with the Baseten Inference Stack.
The company built multi-cloud capacity management (MCM) for over 10 clouds and regions, which it says provides low latency and 99.99% uptime.
AI training teaches models to learn from data, while inference runs models in production. The page compares their hardware, cost, and optimization.
Stories, updates, and other resources from Baseten.
Baseten supports building AI agents with NVIDIA Nemotron 3 Ultra and LangChain Deep Agents, with about 10× lower cost.
A separation of concerns between a control plane and workload planes enables multi-cloud, multi-region model serving and self-hosted inference.
The company charges only for the time a model is actively deploying, scaling up or down, or making predictions. Autoscaling settings can be calibrated to reduce costs.
Serve and scale open-source and custom AI models on the fastest, most reliable inference platform.
The page presents terms and conditions for using the service.
The agreement between the entity accessing or using the service and Baseten Labs, Inc.
Baseten is raising new funding that could push its valuation to approximately $30 billion, sources close to the deal confirm. The AI startup’s latest round would more than double its previous valuation. The funding reflects growing investor confidence in its infrastructure solutions for AI model deployment.
Oct 2026Bloomberg TechNews article
Baseten’s AI inference platform processes over one billion daily requests for customers including Cursor, Notion, and OpenEvidence. The company achieves this scale while supporting real-time AI applications. Baseten has now reached a $13 billion valuation, reflecting strong demand for its infrastructure solutions.
Oct 2026aimagazine.comNews article
Inception AI Inc. introduces **Mercury 3**, a diffusion-based language model that outperforms OpenAI’s **GPT-6 Luna** and Anthropic’s **Claude Haiku 5.5** in speed, achieving up to **8x faster token generation** (1,068 tokens/sec vs. ~240 for Haiku 5.5). The model excels in coding tasks like **SWE-Bench Atlas** (51.1 vs. 33.3 for GPT-6) but trails in long-context tests. Early adopters—such as **Octen AI** (real-time search) and **RevisionDojo** (tutoring)—report sub-second response times, addressing latency-sensitive use cases. Pricing starts at **$0.20M input/$0.70M output tokens**, with API compatibility via OpenRouter, Baseten, and Vercel. Founded in 2024, Inception raised **$50M** (led by Menlo Ventures) and was reportedly in **acquisition talks with Microsoft** (valuation >$1B).
Oct 2026SiliconANGLENews article
Reflection AI released **Beam**, a 501-billion-parameter open model, designed to complete tasks using fewer tokens than comparable models, reducing per-task costs. The company trained Beam with reinforcement learning, penalizing unnecessary token usage while ensuring task completion. Beam outperforms competitors like DeepSeek’s V4.1 Flash in token efficiency, according to benchmarking firm Artificial Analysis. While Chinese models dominate performance rankings, U.S. firms like Pinterest, DoorDash, and Coinbase have adopted them for cost savings. Reflection positions Beam as a lean alternative, targeting enterprise use cases like coding and agent workflows.
Oct 2026VentureBeatNews article
