AI Provider Cost Updates 2026-10-30 | AI Economics News
OpenAI
Run up to 400 questions per request with the now generally available Decisions API
Good news for anyone watching prompt spend: OpenAI made the Decisions API generally available with gpt-6-luna.
The update highlights lower-cost decision workflows, so it’s a good one to look at if you’re sizing up how much your decision-style calls cost.
On top of that, the API supports shared input caching to reduce prompt costs. Teams that send the same context with many questions can benefit from this.
It also expands request capacity, with up to 400 questions per request.
GPT-6.1 Terra is now available in ChatGPT Work, Codex, and the API
Got a new model to evaluate? OpenAI introduced GPT-6.1 Terra as a fast and capable model for computer use, and it’s now rolling out in ChatGPT Work, Codex, and the API.
OpenAI describes it as a fast, efficient model for visual reasoning, computer use, and natural writing. New model launches can shift where workloads run and change cost-performance tradeoffs, so this one is worth a look.
Since the rollout covers Plus, Pro, Business, Enterprise, and Edu plans, more teams get more model choice. Overall usage limits stay unchanged, and workspaces get model-access controls, so you can decide who gets to use it.
Get deeper usage visibility with new Admin Console analytics for ChatGPT Enterprise and Edu
Want a clearer picture of who’s using what? OpenAI expanded ChatGPT Enterprise and Edu analytics across Usage and Insights in the Admin Console.
The update adds workplace health tracking and by-user usage exports. That means you can pull usage data at the individual level for your own analysis.
There are also separate reports for GPTs and Projects. Together, these updates give you better visibility and governance over how your workspace is being used.
Enterprise workspaces can no longer create new custom GPTs
Heads up on a workspace policy change. OpenAI disabled new custom GPT creation for affected Enterprise workspaces.
There’s an exception for FedRAMP workspaces. Existing GPTs remain usable until retirement, so current workflows won’t stop right away.
Because no new GPTs can be created in affected workspaces, this may also reduce ongoing management overhead.
OpenAI and ServiceNow team up to bring models and agents into enterprise workflows
If your company runs on ServiceNow, this one’s for you. OpenAI and ServiceNow expanded their partnership to bring OpenAI models and agents into ServiceNow workflows.
It also goes the other way, with ServiceNow workflows coming into ChatGPT and Codex.
It’s a major provider update with broad enterprise workflow impact. It can affect automation efficiency and AI usage patterns at scale, so it’s one to keep on your radar.
Oracle business apps and AI agents are coming to ChatGPT
Big enterprise news here. OpenAI expanded its partnership with Oracle to bring business applications and AI agents into ChatGPT.
The partnership also powers new Oracle offerings.
It’s a major provider update with broad enterprise impact. It may affect deployment patterns, integration costs, and workload efficiency across large organizations.
Anthropic
Claude Code now summarizes idle long conversations to make the most of an expired prompt cache
Long sessions just got friendlier to your token budget. Claude Code v2.1.314 adds idle compaction across desktop, cloud, Bedrock, Vertex, Foundry, and LLM gateway sessions.
Idle conversations of 200K+ tokens get summarized. The summary is then reused when the prompt cache expires.
The release also improves memory use in long headless and SDK turns. On top of that, it fixes several model-switching and model-mapping issues that affect cost and context-window usage.
Claude Code improves usage-limit messaging and OpenTelemetry reporting
Clearer limits make usage easier to track. Claude Code v2.1.312 improves model handling, usage-limit messaging, and OpenTelemetry reporting.
Usage-limit windows are now clearer. Telemetry is also better around effort levels and MCP servers.
Besides that, the update adds cloud-session workflow improvements and monitoring-related changes. These support observability and operational efficiency.
Claude Code adds a Monitor tool and plugin monitors on Bedrock, Vertex, and Foundry
Want more observability in managed deployments? Claude Code v2.1.310 and v2.1.311 both introduce the Monitor tool and plugin monitors on Bedrock, Vertex, and Foundry.
Version 2.1.310 also brings cloud-session admin controls. It includes efficiency-related fixes for sessions, retries, and usage reporting.
Version 2.1.311 improves memory use for non-interactive sessions during compaction. It also includes startup and session-efficiency improvements that can reduce overhead in large-scale usage.
Gemini Enterprise now supports the Brazil region
Planning where your workloads live? Google added Brazil region support for Gemini Enterprise.
It comes with in-country data residency and machine learning processing.
While the change is regional, it’s a significant global expansion that affects a major market. It may matter for compliance-driven workload placement and efficiency planning.
Gemini Enterprise adds private network connectivity and static IP egress for custom MCP servers
Tighter network control is now generally available. Google Cloud made VPC connectivity and static IP egress for custom MCP server data stores generally available in Gemini Enterprise.
Tool traffic can be routed through Agent Gateway, PSC, and Cloud NAT.
This can improve security posture and network control for enterprise workloads, especially in multi-region deployments.
xAI
Grok Bot can now find unused subscriptions and watch subscription prices
Cutting recurring spend just got easier. Grok Bot 0.78.1 adds shopping cards and subscription-focused templates.
The templates can help you find unused subscriptions, watch subscription prices, and cancel plans with user confirmation.
That helps you identify and reduce SaaS waste and track recurring spend.
Grok Bot adds route-traffic usage visibility and higher routine execution limits
Spend awareness is getting a boost. Grok Bot 0.75.5 introduces route-traffic usage visibility, which helps with spend awareness and operational control in connected workflows.
The release also adds a Vault for saved logins, cards, and contact info, plus email forwarding and higher routine execution limits.
Grok Bot adds Vault payment controls and enterprise audit logs for trusted CA certificate changes
More governance visibility is on the way. Grok Bot 0.77.0 adds Vault payment controls and improves plugin and connector management.
It also introduces enterprise audit log support for trusted CA certificate changes.
Together with major workflow improvements, these changes give enterprises better governance visibility.
Grok Bot trims approval friction with smarter auto-review and scheduling
Fewer unnecessary approvals mean a smoother workflow. Grok Bot 0.75.8 improves auto-review behavior, connector handling, and scheduling logic.
This reduces unnecessary approval friction and improves workflow efficiency.
The update also refines Slack and plugin handling. That can help streamline enterprise usage and reduce operational overhead.
