We've lowered the price of prompt cache reads on Claude Sonnet 5.5 from $0.20 USD to $0.10 USD per million
We've lowered the price of prompt cache reads on Claude Sonnet 5.5 from $0.20 USD to $0.10 USD per million tokens: 0.05x the base input price instead of 0.1x. Cache writes and all other prices are unchanged. See Prompt caching pricing.
Also shipped Oct 7, 2026Anthropic in October 2026 →
The microVM sandbox type is now Generally Available (GA) with GKE Sandbox in clusters that run version...
The microVM sandbox type is now Generally Available (GA) with GKE Sandbox in clusters that run version 1.37.0-gke.4713000 and later. MicroVM sandboxes provide hardware virtualization and isolation for untrusted workloads, AI agent runtimes, and multi-tenant environments.
SSH for Cloud Run services and instances is in Preview
SSH for Cloud Run services and instances is in Preview . Use this feature to establish a secure, interactive shell connection to your running instances.
Set credit usage limits and alerts for your workspace
Workspace admins and owners on paid plans can now set usage limits and alerts that keep shared credits under control. Set a credit threshold for the whole workspace, a project, a member, or, on Business and Enterprise plans, a group or an access token, and choose what happens when usage reaches it: an email or in-app alert, or a block that stops building or...
Faster inline text edits
Simple text changes you make with Edit text inline in the preview toolbar now apply in a few seconds. Text that your app builds from data or translates still takes a little longer.
Gemini 3.7 Flash is deprecated for your app's AI features
Google is retiring Gemini 3.7 Flash ( google/gemini-3.7-flash ), so it is now marked deprecated for your app's AI features and stops working on January 28, 2027. Apps that use it keep working until then, and Lovable no longer picks it for new AI features.
The Cost Analysis tab of the Billing page now includes Serverless Inference Live Usage, which lists the...
The Cost Analysis tab of the Billing page now includes Serverless Inference Live Usage, which lists the estimated cost and token counts of your Serverless Inference requests by model and model access key, for both teams and organizations. …
More from Claude PlatformEvery Claude Platform release →
We've launched Claude Haiku 5.5 (claude-haiku-5-5), our most capable model tuned for high-volume and
We've launched Claude Haiku 5.5 (claude-haiku-5-5), our most capable model tuned for high-volume and latency-sensitive work. It has a 1M token context window, 128k max output tokens, and adaptive thinking with the effort parameter.
Code written for Claude Haiku 4.5 can break on Claude Haiku 5.5
Code written for Claude Haiku 4.5 can break on Claude Haiku 5.5. Manual extended thinking (budget_tokens) returns a 400 error, and adaptive thinking is on by default, so a response can begin with thinking blocks. The same text also counts as more tokens. See the migration guide. For model-specific prompting patterns, see Prompting Claude Haiku 5.5.
The Python and TypeScript SDKs now include classes, in beta, for the browser use tool and the computer use
The Python and TypeScript SDKs now include classes, in beta, for the browser use tool and the computer use tool. You subclass one and write one method per tool against your own browser or desktop automation. The SDK runs the tool loop, the URL and file policies you set for the browser, and your approval callback.
Claude Max and Team plans now include monthly API credits
Claude Max and Team plans now include monthly API credits. To learn how to claim them, see API credits for Max and Team plans.
In Claude Managed Agents, a cloud environment with limited networking now also applies its allowed_hosts to
In Claude Managed Agents, a cloud environment with limited networking now also applies its allowed_hosts to the web_search and web_fetch tools. A web_fetch call for a URL on a host that allowed_hosts does not match returns a url_not_allowed error result to the agent. web_search omits results from such hosts.
With limited networking, creating a session fails with a 400 error when an enabled web tool's allowed_domains
With limited networking, creating a session fails with a 400 error when an enabled web tool's allowed_domains has an entry not within allowed_hosts. So does a session update that adds such an entry. An allowed_hosts entry matches one exact host unless it starts with *., so docs.example.com is not within ["example.com"].
In Claude Managed Agents, the web_fetch tool now fetches only URLs that have already appeared in the session
In Claude Managed Agents, the web_fetch tool now fetches only URLs that have already appeared in the session, for example in the text of a user message, in a web_search result, or in a page that web_fetch returned earlier. This reduces the risk of data exfiltration.
The Compliance API chat endpoints now also return chats from the unified Claude experience, in beta for
The Compliance API chat endpoints now also return chats from the unified Claude experience, in beta for Claude Enterprise organizations, with your existing Compliance Access Key. See Retrieve and delete chats, files, and projects.