A/B test models in production
Shadow traffic proves a candidate is operationally sound. It can't tell you if users like it better. Run the split at the endpoint instead of in your app code.
// Company Updates
All companiesFrom the Industry Leaders
DeepSeek V4 Pro 0813 vs Claude Fable 5 on DeepSWE: Cost, Coding, and Routing
We ran 904 DeepSWE rollouts on DeepSeek V4 Pro 0813 and Claude Fable 5. Fable leads pass@1 at 90x the cost; Pro wins pass@4, and a Pro-first cascade hits 82.7%.
A Preview of DuckDB v2.0
DuckDB v2.0 is coming this fall. In this post, we preview its headline features: DuckDB as a server, triggers, the VARIANT type, asynchronous I/O, a new SQL parser, a new storage format, and much...
GPT-5.6 Sol is 50% off on AI Gateway for the next month
GPT-5.6 Sol, the flagship of OpenAI's GPT-5.6 series, is 50% off on AI Gateway through September 18. The discount applies on the OpenAI provider to all token types, tiers, regions, and modes, and it...
Deploy Cursor Origin repositories with Vercel in public beta
Pro customers can now connect Origin repositories to Vercel. Once connected, pull requests on Origin will automatically generate Preview deployments on Vercel, and merging will trigger a Production...
Distributed Layerwise Offload: Scaling Toward 200B+ DiT Models Efficiently in vLLM-Omni
Distributed Layerwise Offload shards and streams DiT weights across devices, serving a measured 124 GB Cosmos3 model on 64 GB HBM and estimating a path toward 200B+ models.
Grok 4.6 in GitHub Copilot
Grok 4.6, our latest coding model, is now available in GitHub Copilot.
Using AI_Functions in Your Data Warehouse: Top Use Cases
In most organizations, data warehouses hold structured data, while unstructured data...
Why agentics projects fail and how to fix them
Why do some AI deployments yield great successes while so many still crash and burn? The answer is in the data.
How dbt State cuts warehouse compute and speeds up every run
How Fanatics cut warehouse compute by only rebuilding what's changed
Universitas Gadjah Mada, Indosat and NVIDIA Open Indonesia’s First University AI Center to Develop Local AI Talent
Indonesia is taking charge of its AI future. This week, the Ministry of Communication and Digital Affairs (Komdigi), Indosat Ooredoo Hutchison (Indosat or IOH), NVIDIA and Universitas Gadjah Mada...
What is Query Caching?
Query caching stores the results of previously executed queries so that identical or compatible future …
Custom reward functions for multi-turn reinforcement learning with Amazon Nova Forge
In multi-turn reinforcement learning, your custom reward function decides what the model actually learns. This post shows how to design a composite multi-turn reward for Amazon Nova Forge, execute...
Encrypted Client Hello (ECH) is now supported on Vercel CDN
Vercel CDN now supports Encrypted Client Hello (ECH) for domains managed by Vercel DNS. ECH encrypts the Server Name Indication (SNI) in the TLS handshake, the last part of an HTTPS connection that...
Zero-Copy Data Architecture: Snowflake Summit 2026
Explore how zero-copy data architecture and enterprise AI help leaders reduce costs, analyze unstructured data and make faster decisions.
Building agentic workflows with SageMaker AI and Bedrock AgentCore
Learn how to combine OpenAI-compatible endpoints on Amazon SageMaker AI with Amazon Bedrock AgentCore runtime to build a multi-agent workflow where each specialized agent uses the model best suited...
Neon skills landed in the Vercel CLI
Vercel just gave their CLI a new trick: when you install a Marketplace integration from the terminal, it also installs that provider's agent skills. A single command now provisions a Neon database on...
Life Sciences M&A Data Integration with Snowflake
Learn how Snowflake accelerates life sciences M&A and divestitures with secure data sharing, clean rooms, governance, and faster integration.
How Neuron Systems Served 2.3 Million Fans Across 104 World Cup Matches with AI on Confluent
Every one of the 42.1 million production events flowed through Confluent’s Data Streaming Platform and that single decision is what let a small team serve a global tournament without the commentary...
Adaptive Verification in vLLM: DSpark confidence-scheduled verification
Sizing the DSpark draft-verification budget from per-request confidence instead of verifying every drafted token, so one configuration holds the throughput/latency frontier from batch size 1 to 256.