Topic
Assigning tasks across models by cost, speed and quality instead of routing all work to one large model.
Bank of America will double its AI budget next year, CEO Brian Moynihan said; about 140 uses now cost $400 million and yield $800 million in benefit. Its technology chief, Hari Gopalkrishnan, says every AI project goes through a review covering 16 categories of risk.
Reality
Cloudflare's Auto Router, now in public beta, picks a model per AI Gateway request and cut costs by up to 30% in the company's own OpenCode use. How much of that transfers depends on how much of a team's traffic never needed a frontier model.
Reality
CData's Connect AI Gateway, now in early access, holds AI agents to each user's own permissions down to specific rows and columns. Teams deploying agents can now buy that control, though the accuracy claims behind it come from CData's own tests.
Reality
LLM Gateway's own benchmark found Smart Routing cost 33.2% less than a premium model and 35.6 times as much as a low-cost one that stayed competitive. With the premium saving not statistically established on 80 prompts, a fixed cheap model is the baseline Smart Routing has to beat.
Reality
Salesforce says many of the technologies behind its new Enterprise AI Harness already run in its cloud services, and the rest arrive early in fiscal 2028. It has not said which features sit in which bucket.
Reality
The gap implies each session now uses roughly a quarter fewer requests, so the compaction threshold and the prompt cache in Uber's own harness are carrying part of a saving usually credited to cheaper open models.
Reality
The router picks a cheaper model for work that resembles work it has already seen, and falls back to the strongest one when nothing matches. The same traces that enable that call are the proposed training set.
Reality
Auto now picks the model for each task in every Replit account, and getting a specific model back means a Core or Pro plan plus a mode that can charge usage credits. Stripe's reported $8 billion for OpenRouter prices the same layer.
Reality
A dev.to hands-on replaces the classifier Lambda with a Step Functions Choice state. The trade is a second Bedrock invocation on every question, including the cheap ones.
Reality
A dev.to walkthrough moves Claude, Nova and Llama IDs out of Lambda code and into AWS AppConfig feature flags. Rollback stops being a deploy, and model choice stops being a code review.
Reality
A single deep-research request burned 194% of a rolling quota, most of it inside six minutes. The tiering rule that should have caught it never reached a decision point.
Reality
Glean says its customers mostly turn on automatic model selection to control spend, not to improve answers. The routing layer, not the model, is where enterprise AI budgets now get decided.
Reality
Foundry, announced August 17th, lets brokers describe a workflow in plain English and deploy it to one lane. The buy decision moves from which agents ship to who owns the process logic.
Reality