Cursor

Cursor confirmed its acquisition by SpaceX has closed. The team joins SpaceXAI to work on Grok, Grok Build, Grok Bot and the Grok API alongside Cursor itself, putting a widely used AI editor under the same roof as the Grok model line.

Cursor has launched Origin, its own code hosting platform, now rolling out in beta and wired directly into the editor. Teams get started by syncing existing repositories over from GitHub.

Cursor had a team of agents rebuild SQLite from its 835-page manual, producing a Rust replica that passed a full held-out test suite, with cost varying 15x depending on the model mix.

Cursor has released new plugins that let its AI agents read, write, and take actions across Gmail, Google Drive, Calendar, Docs, and Sheets. This extends Cursor's agentic capabilities beyond code editing into everyday productivity and workspace automation tasks.

Cursor reports Fable 5.1 at 73.4% at max effort on CursorBench 3.2, the highest it has measured for any model on that harness, and credits the model's habit of verifying its own work for carrying harder tasks end to end. Available in Cursor now.

Claude Opus 5 lands in Cursor, matching Fable 5 on CursorBench at half the price and, unlike Fable, compatible with Zero Data Retention.

Cursor’s iPad app extends its mobile coding workflow with more room for agents plus an inbox and pull-request review covering comments, checks and approvals.

Cursor announces a greenfield flow: create a new web app in the editor, store the code with Origin, and deploy to Vercel. The announcement is brief and the demo video was not transcribed, so details of Origin and limits are not covered.

Cursor added Gemini 3.8 Flash to its model list on release day and published its own CursorBench numbers for the model, giving a side by side against the other coding models Cursor offers.

Cursor reworked how its cloud agents handle MCP connections, skills, and computer-use actions, cutting token consumption 20-30% on typical runs and up to 80% where computer use dominates — shifting the ceiling on how ambitious a delegated task can be per budget.

Cursor's cloud agents can start from events, keep a long-lived goal, monitor PRs and Slack threads, and run scheduled tasks. Subagents get isolated VMs with their own project copy, skills can be pinned as modes, and steering now waits for the next tool call. Demo videos were not read.

Cursor now supports Agent Plugins, an open standard for bundling skills and MCP servers for use across agents. https://t.co/avJch1Kblq

Cursor adds GLM 5.3 and GLM 5.3 Flash, and reports GLM 5.3 Max as the best-scoring open-weight model on its CursorBench 4.0.

Cursor's cloud agents now start three times faster, using development environments prepared continuously in the background at no extra cost.

Cursor now prepares development environments continuously in the background, cutting cloud-agent start times about threefold at no extra cost. A build that fails never ships — agents keep working from the last successful one while you debug it separately.

Cursor added DeepSeek V4.1 Flash to its model list. Cursor reports every effort level on its CursorBench cost frontier at $0.60 to $0.98 per task, with no higher-scoring model at that cost. Enable it in Settings under Models.

Cursor adds Meta Superintelligence Labs' Muse Wonder as a selectable model. Cursor reports 46.1% on CursorBench at default effort, its best Muse result, with pricing of $4 per million input and $20 per million output tokens. Only the post was available, not the linked page.

Our hope is that this lowers the barrier to AI research, so more labs can train models efficiently. Here's how we built it: https://t.co/EaC2z7Mw3a

Cursor released its Agent SDK. Agents are defined as code, with prompts in Markdown and tools in TypeScript, deployable on Cursor's hosted infrastructure or self-run, with built-in evals to measure each change.

Cursor Router is available today across all surfaces on Teams and Enterprise plans. Admins can enable Router per team, allow or block certain models, set defaults, and disable select optimization modes for users. Learn more: https://t.co/g7T7cHwRZx

Cursor ships a /visualize command in its Agents Window that analyzes data and renders charts and diagrams inline in the chat. The announcement is brief, and the demo video could not be read.

Read more about how we set up our cloud agent environment: https://t.co/sVSv1HpBn9

Cursor now hosts code. Origin's early beta ships repos, pull requests, browsing and two-way GitHub sync, with agents able to answer questions, push branches and update PRs in place. Vercel, Depot and Buildkite connect for previews and CI. Live on all paid plans, enterprise admins can opt out.

Cursor learns model choice from live traffic rather than benchmarks: Compass first predicts whether a turn is simple enough for a price-efficient model, then a learned taxonomy of tasks, domains and modifiers picks the frontier model. Labels come from what the user did next; costs count cache misses