Concentrate AI Changelog, Nov 16, 2026: Spend Alerts, BYOK
Concentrate AI changelog for the week of Nov 16, 2026: spend limit alerts on any team, member or API key, BYOK ordering, API key rotation and 9 new models.
Concentrate.ai
On this page
The Concentrate AI changelog for the week of November 16, 2026 adds spend limit alerts on any team, member or API key, forecast alerts, BYOK ordering, API key rotation and 9 new models. Concentrate AI is an LLM gateway: one API for models from every major provider, with spend limits by team, member and key. Every change is listed in the Concentrate changelog.
Alerts: spend limit and forecast alerts for teams, members and keys
Alerts can now be set on your organization, a team, a member, an API key or your personal account.
Spend limit alerts on any team, member or API key. Pick who gets notified for each alert. Alerts go out by email, SMS or webhook. See key spend limits.
Forecast alerts. Limit exhaustion alerts (previously called Key exhaustion) now forecast when a key, team, member, organization or personal account is on track to hit its spend limit.
Enforced alerts. An organization or team can enforce an alert so it applies to every team, member and key below it.
Entity limit alerts can fire at a fixed amount remaining, and some alerts can run every minute. Alerts are also available in the API under settings.config.alerts.
LLM Gateway
Instant access to every model
Change models, providers, add fallbacks, and more by running all AI usage through one platform.
API: 100 MB images, tool_usage and more sampling parameters
Images up to 100 MB. Chat requests can now be up to 101 MiB.
tool_usage in responses. Every response now includes the OpenAI tool_usage object with the number of web searches billed. usage.tool_calls is deprecated. See the web search docs.
More sampling parameters. frequency_penalty, presence_penalty, logit_bias and seed are passed through to models that support them.
Dashboard: a page for every key, team and member
A page for every key, team and member. Each one has Overview, Usage, Keys, Teams, Settings and Alerts tabs, as they apply.
Member settings. Set limits and other settings for an individual member of an organization or team.
Redesigned settings. Settings for keys, teams, members and organizations use a new tile layout for limits, rate limits, routing and data retention.
Logs and model picker. Filter logs by deleted API keys. The model picker opens scrolled to your current model.
Web searches on streamed OpenAI responses are counted correctly.
Streamed /v1/messages content blocks are numbered in order, and thinking blocks open before their reasoning content.
Setting reasoning.summary returns summarized thinking from Claude models.
Low balance emails and SMS show your current balance.
An entity limit alert fires again after you change the limit.
The Windows setup scripts download without a login redirect, and the OpenCode setup stops if OpenCode is not installed.
In HIPAA scopes, the dashboard assistant no longer looks up public docs.
FAQ: Concentrate AI changelog, week of November 16, 2026
What is new in Concentrate AI this week? The week of November 16, 2026 added spend limit alerts on any team, member or API key, forecast alerts, enforced alerts, BYOK ordering, API key rotation, requests up to 101 MiB and 9 new models. The full list is in the changelog.
Can I set a spend alert on one team, member or API key? Yes. In Concentrate you can set an alert on your organization, a team, a member, an API key or your personal account, and choose who is notified by email, SMS or webhook.
Can Concentrate warn me before a key hits its spend limit? Yes. Limit exhaustion alerts forecast when a key, team, member, organization or personal account is on track to hit its spend limit.
Can I use Concentrate capacity first and my own provider keys as fallbacks? Yes. With BYOK you can use Concentrate capacity first and keep your own keys as fallbacks, and you can add more than one key per provider and set the order they are tried.
How do I rotate a Concentrate API key? Open the keys table in the dashboard and choose rotate from the actions column for that key.
Which models did Concentrate add this week? Claude Fable 5.2, Gemini 4 Argon, Gemini 4 Carbon, Kimi K2.7 Code, Qwen 3.8, Qwen3.6, HY4, Nemotron 3 Nano and Gemma 3 12B. Each one is in the model catalog.
What is the largest image I can send through Concentrate? 100 MB. Chat requests can be up to 101 MiB.