Compute
Dell says 80% of AI tokens come from open models
Dell says that the high cost of inference moves AI work to systems that companies control.
siliconangle.comClaimed, not confirmed
Tag
Nvidia
68 stories carry this tag.
Models
Reflection releases Beam, an open model with 501 billion parameters
Beam has a score of 30 on a test by Artificial Analysis, and uses a small number of output tokens.
Models
Microsoft adds 4 open models and shared hosting to Foundry
Teams can tune 4 open models in Foundry, and many tuned adapters can share one parent model.
Models
Hugging Face gives out Carbon-X, a DNA model for a laptop
Carbon-X has 14 billion parameters and answers questions about DNA in plain words, but nobody tested its accuracy in a lab.
wired.comClaimed, not confirmed
Models
Nvidia shows open models for document search at GTC Berlin
Nvidia tells that its 8 billion parameter models are first on a test, but only Nvidia gives this result.
theregister.comClaimed, not confirmed
Industry
Mercor says the average pay of its AI experts is $110 an hour
Mercor gave no data on the share of tasks that pay $110, and the work can end fast.
Deals
doubleAI gets $200M for Maxwell, a model for fast code
The company says Maxwell made one data tool 38 times as fast, but it gave no external test.
siliconangle.comClaimed, not confirmed
Security
ServiceNow adds AI Passport to set limits for AI agents
In a demonstration, the agent sends an $18 refund but the system stops a $40 refund at runtime.