npx @buildinternet/releases get neonRecently shipped the Neon backend GA — Postgres, Object Storage, Functions, Managed Better Auth, and AI Gateway, all branching with your data.
The Neon backend is GA. Every branch is a full copy, so a PR, preview, or agent session gets the whole backend, not just the database.
DATABASE_URL and credentials injected.Free plan storage doubled. Database storage is now 1 GB per project, up from 0.5 GB, across 100 projects. Existing projects pick up the new limit automatically.
AI Gateway dropped its access requests. On paid plans with prepaid credits, every model in the catalog is callable — OpenAI, Anthropic, Google, Meta, Moonshot — and Claude Opus 5.5 joined the list. Embedding models are supported too, so retrieval pipelines embed on the branch instead of calling out.
Functions gained triggers and custom domains. Function Triggers now fire on Object Storage uploads and on schedules, so backend jobs no longer need an external scheduler or pg_cron with scale-to-zero disabled. Functions can also sit behind a custom domain like api.example.com, at no extra charge.
**
Stripe Data Pipeline now supports real-time sync to Postgres, streaming Stripe API data into a Neon database without custom API or webhook integrations; Stripe manages backfill and ongoing delivery with no rate limits. Neon is a connectable destination from the Stripe Dashboard, and the feature is in public preview and free to start.
Neon Realtime turns any SQL query into a live query that stays up to date as data changes, running from Lakebase Postgres with scale-to-zero, branching, and type-safe end-to-end reactivity, in public beta across four AWS regions. Also adds 11 models to the AI Gateway, real-time Stripe data sync into Postgres, CJK full-text search in Lakebase Search, and Managed Better Auth on Better Auth 1.7.6.
Neon Realtime is live in public beta, letting any deterministic SQL query run as a live query that stays up to date as data changes, with no polling or refetch logic. It runs inside Lakebase Postgres rather than a separate service, and comes with a TypeScript SDK with subscribe() and useLiveQuery, SSR support, instant local writes, and agent setup via the Neon MCP server or the neon-realtime skill.
Every model in the Neon AI Gateway catalog, including Claude Opus 5.5, is now callable on paid plans with prepaid credits without requesting access. Feedback can be sent via the new send_feedback MCP tool or neon feedback CLI command, and neon deploy and neon config apply now accept -y.
Neon's AI Gateway now exposes POST /v1/embeddings on the same OpenAI-compatible /v1 path and credential used for chat, accepting a single string or a batch of up to 150 strings, with qwen3-embedding-0-6b (1024 dimensions, configurable) and gte-large-en (1024 dimensions, cosine distance) available at launch. Vectors can be written into Lakebase Postgres and queried with Lakebase Search, keeping the full retrieval pipeline inside one Neon branch.
Neon Free plan projects now get 1 GB of Postgres storage each, up from 0.5 GB, with existing projects updated automatically and the 100-project limit unchanged. Each project also includes 100 CU-hours of compute per month, autoscaling to 2 CU, 10 branches, a 6-hour instant restore window, Managed Better Auth for up to 60,000 MAU, 5 GB of S3-compatible object storage, and Node.js functions with 1M monthly invocations.
Neon Free plan projects now get 1 GB of database storage, up from 0.5 GB, with existing projects picking up the new limit automatically. The Neon AI Gateway also now supports embedding models via /v1/embeddings, returning 1024-dimension vectors billed on input tokens only.
@neon/effect wraps the Neon SDK's client with the same method names and parameters, returning API calls as Effects and paginated lists as Streams; SDK errors become tagged errors handleable with Effect.catchTag. It requires Effect 4 and installs via npm install @neon/effect effect.
Neon outlines reliability practices adopted after two control-plane outages in one week, including postmortems that lead with customer impact and quantify detection and mitigation times, plus a rule to halve each response interval. The post also covers AI-assisted impact measurement, where agents must produce rerunnable notebooks pairing each figure with its query.
Lakebase Search is generally available, adding the lakebase_vector extension for semantic search and lakebase_text for BM25 full-text search inside Lakebase Postgres. The vector index lives in object storage and scales to zero without rebuilding, and pgvector users can add a lakebase_ann index without migrating data or changing queries.
Neon details how Lakebase Postgres's decoupled compute-and-storage architecture makes restores a metadata operation that creates a branch at a timestamp, dropping recovery time to seconds even for 100 TB databases instead of the multi-hour provisioning, snapshot hydration, and WAL replay of traditional managed OLTP. The post cites a survey of 50 developers running production Postgres in which 59% had a critical production failure in the past 12 months and only 21% recovered in under 60 minutes.
Reset from parent now rolls back a branch's Object Storage buckets and objects to the parent's current state along with its Postgres data, reverting uploads, overwrites, and deletes made on the branch, in AWS US East (Ohio) and AWS Europe (Frankfurt). Lakebase Search extensions lakebase_vector 1.1.1 and lakebase_text 0.1.3 ship performance improvements and fixes.
Interview with Electric co-founder James Arthur describes plans to build a next-generation realtime sync service natively into Neon and Lakebase, with native support for scale-to-zero and branching. The team aims to solve the expressiveness-versus-performance tension that makes realtime sync fail in production on Postgres, following Neon's backend GA.
A new storage_object_created trigger invokes a Neon Function when an object is created in an Object Storage bucket, passing the bucket name and object key in the POST body and supporting an optional key prefix to scope it to a path. Functions are long-running and remain compatible with scale to zero, and the schedule trigger is also available.
Neon Functions now support custom domains, replacing the Neon invocation URL with a branded hostname registered from the Console, CLI, SDK, or API. Domains are attached to one Function on one branch, and TLS certificates are provisioned automatically through Let's Encrypt.
Function Triggers let a branch-scoped cron schedule invoke a deployed Neon Function, keeping the timer outside the Postgres compute so scale to zero can stay on, unlike pg_cron. Triggers run JavaScript or TypeScript that can call HTTP, Object Storage, and the AI Gateway, and child branches inherit parent triggers disabled until enabled.
Neon's backend is now generally available, extending branching beyond Postgres so every PR, preview, or agent session gets a full copy of the schema, object storage, auth, and functions. The release adds Function Triggers for schedules and object-created events, plus a declarative neon.ts config applied with neon deploy.
Claimable Neon lets agents provision a temporary project with scoped credentials without an account or payment details, with humans claiming ownership later via neon claim accept; Lakebase Postgres, the Data API, and Managed Better Auth are supported today. The new @neon/tools package exposes typed function-calling tools from the @neon/sdk with adapters for MCP, Mastra, and Eve, adding over 70 tools to the hosted Neon MCP server, and large fixed-size computes of 18 CU and above now expand shared buffers to use up to 75% of available memory.
Claimable Neon lets AI agents provision temporary Neon projects using anonymous registration without creating an account, then humans can sign in later and claim the project into their organization. The feature implements the auth.md protocol, supports Postgres, Data API, and Managed Better Auth, and uses 72-hour expiration with 100 MB storage and 1 GB transfer caps for unclaimed projects.
Neon now sizes Postgres shared_buffers to 75% of the compute instance's memory capacity automatically, making the default more robust than a fixed 128 MB. The capacity-aware default applies across compute profilesholones; for most workloads it requires no configuration, though specific tuning guidance applies to multi-instance deployments.