Changelog 2026-09

AI Tasks, Security, and Stability Updates

Feature Updates

• Agent Computer maintenance controls added: Admins can pause new sessions, notify users before submission, and optionally let existing conversations finish.
• Model tiers added for daily tasks: Agents can select a persistent model tier to balance quality, speed, and cost.
• Event-triggered task titles and reasons added: Chats receive event-based titles and display launch reasons without overwriting user-edited titles.
• Welcome credits added for free workspaces: Free workspaces now receive 200 one-time credits instead of 200 monthly credits.
• Custom domain grace period added: Business plan domains remain active for one week, with expiration notifications and uninterrupted access upon renewal.

Fixes & Improvements

• Strengthened sign-in security: Account lockout is now enabled by default; administrators can disable it with SIGNIN_ACCOUNT_LOCKOUT_ENABLED=false.
• Fixed chained formula calculations: Blank text remains blank in numeric calculations instead of becoming 0, preventing incorrect results.
• Improved deleted record cleanup: Comments on deleted records are inaccessible and permanently removed with related subscriptions.
• Fixed incomplete shared view fields: Shared grids and expanded records now show all visible fields, including unscrolled columns.
• Improved multi-role permissions: Editing, record visibility, and field permissions now remain consistent across combined space, department, and Base roles.
• Improved permission-aware search: Searches respect filters and row permissions; unsafe large-table searches pause until searchable fields are configured.
• Fixed inconsistent large-number display: Large numbers now remain consistent across cells, record details, copied values, and API responses.
• Improved rollup refresh reliability: Relationship changes recalculate rollups for previous and new linked records, including dependencies after relationship conversion.
• Improved manual sorting in large grids: Row order remains stable during concurrent edits, recovery, and view or table duplication.
• Fixed table deletion failures: Tables no longer become hidden without deletion, and recovery is more reliable after interruptions.
• Fixed sidebar table ordering: Table positions persist after refresh; failed tables are hidden while import errors remain accessible.
• Fixed nested grouping freezes: Group statistics with date formulas are more stable, and unrelated results no longer flicker.
• Fixed temporary linked-record titles: Titled linked records no longer briefly display “Untitled” after bulk updates.
• Improved large-file imports: Large CSV and Excel imports now fail less often and produce fewer temporary request errors.
• Fixed blank SVG thumbnails: SVG attachments now display thumbnails correctly in grid views.
• Improved AI chat queues and recovery: Queued messages show status, rejected messages disappear, and ordering and reconnection recovery are clearer.
• Improved AI conversations and page loading: Streaming responses are smoother, while Base and space pages load previews only when needed.
• Improved AI field autofill performance: Bulk processing is faster, and existing fields now support workspace-level custom providers correctly.
• Improved App Builder recovery: App Builder recovers more reliably from preview failures and temporary workspace outages.
• Restricted organization department API results: The organization user API returns only users’ departments; integrations can use the dedicated department list API.
• Strengthened account and integration security: Improved security across accounts, licenses, integrations, and sign-ins, while eliminating invalid organization-linking errors.

GPT-6 Sol and Luna Are Now in Teable

Automation, Billing, and Stability Updates

Feature Updates

• Added custom domain renewal grace period: Domains remain active seven days after Business expiration; owners receive expiration and deactivation notices.
• Added connected-app event triggers: Automations and Routines can use event data; copied workflows support reconnection during credential setup.
• Added third-party app tool billing: Calls use provider-priced credits shown in credit history; failures, searches, and blocked calls are free.

Fixes & Improvements

• Improved connector stability: Expired, deleted, or replaced connections show disconnected, notify editors, and retain enabled status for reconnection.
• Improved connector events and Routines: Fixed event discovery, delayed runs, template installation, expired subscriptions, credit checks, and trigger test results.
• Improved lookup and rollup display: Lookup-based rollups show text and preserve numeric and Boolean types in standard and conditional rollups.
• Improved connected app credential stability: Fixed imports failing despite valid authorization and streamlined credential replacement and disconnection.
• Improved billing accuracy for long-running jobs: Prevented completed jobs from being charged repeatedly.
• Improved numeric lookup sorting and grouping: Values now sort numerically, with corrected grouping order for referenced linked records.
• Improved AI session stability: Enhanced model switching and recovery, especially for long-running Routines using non-default models.
• Improved real-time linked record titles: After bulk updates, many-to-one and one-to-one fields show saved titles without refreshing.
• Improved application security and reliability: Strengthened system protection and runtime stability for a more reliable experience.

Connect 1,500+ apps to Teable AI

1. Turn Gmail Inquiries into Sales Leads

2. Bring Attio contacts into Teable

3. Draft, publish, and track LinkedIn posts

X integrations, calculations, and performance optimization

Feature Updates

• Added X OAuth integration support: Connect services like X that require the deployment’s own OAuth app; configure API key integrations through credential forms.

Fixes & Improvements

• Improved chat card interactions: Pending authorization, connection, selection, credential, and question cards no longer block new messages; layouts and actions are clearer.
• Improved cross-table calculation consistency: Chained formulas, lookups, inventory, and allocations now update promptly when linked records change.
• Fixed unexpected scrolling when selecting an Assignee: The field being edited stays visible in record details.
• Improved integration connections: Clearer loading indicators and empty states, plus better mobile controls, make connections easier to understand.
• Improved table operation stability: Reduced false “Table not found” errors during field changes and improved pending or failed imports.
• Improved linked-record deletion performance: Faster, more reliable link removal across required, self-referencing, one-way, two-way, calculated, and recycled relationships.
• Fixed App Builder message queue failures: Queued messages now enter the generation process; previously affected messages must be resent.
• Fixed grid view misalignment during fast scrolling: Previously viewed records no longer briefly appear in the wrong position.
• Improved reconnection for idle tables: Recovery progress is shown, avoiding premature “table unavailable” messages.

• Adjusted Sheet feature availability: Due to compatibility limitations, the Sheet form designer, preview, and sharing panel are unavailable in this release.
• Fixed personal view loading: Personal views no longer occasionally remain blank with a loading placeholder; content appears without refreshing.
• Improved public template redirects: Outdated public template links now redirect to the Teable template page instead of returning a 404 error.
• Improved data operation performance: Faster table, record, field, and SQL query operations improve responsiveness with large datasets.
• Fixed expandable content: Recently measured expandable content now opens properly, improving page interaction stability.
• Improved real-time collaboration performance: Fewer unnecessary refreshes and polling requests during view configuration changes make collaboration smoother.
• Improved table list consistency: The table list reflects structural changes, including field updates, more promptly and accurately.
• Fixed member dropdown scrolling: Member lists in expanded records scroll again, making it easier to find and select members.

Continuous operation, performance, and stability upgrades

Feature Updates

• App Builder save failure guidance: Offers cleanup tips and sends failure details to AI when oversized versions cannot be saved.
• Continuous Routine execution: Routines run until completion, failure, or manual termination; histories now show start and end times.

Fixes & Improvements

• Improved template preview workflow: Creates a new copy and opens settings, matching Template Center; fixes grid record loading.
• Improved App Builder chat startup: Chats open in preview mode by default, with clearer environment clipboard labels.
• More reliable shared Base saving: Fixes failures when linked tables or fields are missing.
• More accurate Date field sorting: Sorts by full timestamps despite hidden times; date groups still follow display formatting.
• Improved bulk collaborator assignment notifications: Reduces duplicate emails, summarizes multiple records, and adds clearer wording and table links.
• Improved account deletion flow: Clearly lists owned spaces and required actions; fixes cleared desktop confirmations and broken mobile links.
• Improved shared chat unread notifications: Unread markers and reply alerts follow the latest questioner, preventing uncleared states for other viewers.
• Better complex Base performance: Speeds up Bases with complex linked records and Lookup calculations for smoother browsing and operations.
• Improved Secret rotation: Preserves existing credential bindings to avoid reconfiguration; new access permissions still require approval.
• Safer email verification code login: Adds security checks before sending or resending codes; related integrations must verify when enabled.
• More reliable app publishing: Optimizes publishing for projects with many bound Secrets, improving speed and success rates.
• Safer Formula saving and calculations: Checks edits before saving; failures remain drafts, preserving valid definitions and values without overwriting results.
• Fixed blank Record History panels: History now displays after changes to Formula, Rollup, and Conditional Rollup fields.
• More accurate audit logs: Distinguishes user-created personal access tokens from automatic token refreshes, helping administrators track actions accurately.
• Reliable external app connections: Fixes Gmail and API Key connections, prevents canceled connections reusing unrelated accounts, and ensures Agent access.

Collect data from 800+ websites with Teable AI

Describe the data you need. Teable Agent scrapes it from 800+ public websites and platforms—including LinkedIn, Amazon, Google Maps, Instagram, TikTok, X (Twitter), and YouTube—all in one place.

1. Find Beauty & Skincare Influencers on YouTube

2. Find Local Businesses on Google Maps

3. Compare Products on Amazon

Image Generation, Account and Experience Upgrades

Feature Updates

• Added queued image generation: Track, cancel, select models, and deliver images to galleries across chat, CLI, App Builder, and routines; pay only for delivered images.

Fixes & Improvements

• Improved account deletion: Unified web flows guide space transfers or trashing; no trash emptying or manual exits required. Subscription safeguards remain.
• Improved .tea project imports: Preserve integration and secret placeholders without credentials for easy authorization; legacy exports remain supported.
• Improved integration and automation management: Manage authorization in the Integrations dialog and enable automation per workflow.
• Improved App Builder previews: Show last-saved previews on startup, preserve usable previews after setup failures, and reduce incorrect stopped/failed statuses.
• Updated media generation APIs: Create, retrieve, and cancel one task per request; batch, continuation, and retry flows are no longer supported.
• Improved image generation reliability: Better handling of failures, rejected requests, stalled tasks, and duplicate results improves output and billing reliability.
• Improved chat readability: Clearer attachments and mentions, contained long content, and collapsed tool/image rows reduce layout shifts; failures remain expandable.
• Improved routine status displays: More accurate execution statuses reduce transient error indicators during loading.
• Improved restricted view visibility: Inaccessible views no longer appear in table views or recently visited lists.
• Improved account and shared-link security: Strengthened account, authentication, and shared-link protections; existing password-protected links may require password re-entry.
• Improved login experience: Returning signed-out users reach the login page; switching authentication tabs preserves entered email addresses and passwords.
• Improved permanent deletion: Permanently deleting tables or bases also removes related comments and comment subscriptions, preventing orphaned content.
• Fixed import entry visibility: The import entry now displays and opens correctly when importing is the only available action.
• Fixed attachment access: Restored Aliyun OSS attachment links while preserving downloads through other supported storage services.
• Fixed imported field references: AI classification and generation fields no longer reference source projects, preventing misleading circular-reference errors when copying/publishing templates.
• Improved favicon clarity: Sharper favicons in browser tabs, link previews, and high-density displays.
• Fixed field-level permissions: Applying permissions no longer incorrectly hides/locks fields or unexpectedly restricts records without field restrictions.

Login, Automation, and Performance Optimization

Feature Updates

• Added email verification code sign-in: After email setup, registered users can sign in with a code for easier access.
• Added Sign in with Apple: Once configured, web sign-in supports Apple alongside Google, GitHub, and OIDC.
• Added Routine authorization scopes: Personal access tokens and OAuth apps can authorize Routine access for easier automation integration.
• Added Routine CLI management: Use the CLI to update, enable, disable, or start Routines within existing permissions.
• Added Routine result display: Run details show the assistant’s final response when available for quick result review.

Fixes & Improvements

• Optimized image uploads: Upload support now matches the selected AI model tier’s capabilities, ensuring compatible image features.
• Fixed App Builder file handling: Resolved file link and download issues, ensuring generated files open in the appropriate previewer.
• Improved automation script editing: Edits persist after closing dialogs, and workflow tests use the latest saved configuration for consistent results.
• Improved notification display: Stacked notifications now remain clearly visible when multiple messages appear simultaneously.
• Fixed AI chat button display: Prevented the send button from being compressed in constrained layouts, improving chat usability.
• Improved Routine’s Chinese experience: Refined Simplified Chinese terminology, editor styling, and form consistency for easier configuration.
• Improved product and API documentation: Replaced “Base” with “Project” and refined API route titles and authentication examples.
• Strengthened app and automation permission isolation: Prevents cross-project, space, integration, or tenant access; workflows using broad inherited permissions require review.
• Improved record query stability: Supports timeouts for slow record, count, and aggregation queries, with clearer errors to prevent resource overuse.
• Optimized record query performance: Skips unnecessary total counts by default while retaining them when required, reducing database load.
• Improved record pagination reliability: Deleting cursor records no longer causes missing or truncated results; invalid or mismatched cursors show errors.
• Fixed multi-value filtering: “Contains” and “does not contain” now work when JSON or multi-value field content includes apostrophes.
• Improved calculated field stability: Enhances calculation reliability and efficiency, clearing calculation warnings only after related fields finish recalculating.
• Optimized view updates: Views restored through undo, redo, or trash appear immediately, with fewer unnecessary table and view refreshes.

Improved right-to-left support

• Layouts that follow your language. Sidebars, menus, and dialogs adapt to right-to-left use.
• Better typing in AI Chat and App Builder. Chat inputs follow your interface language’s text direction.
• Clearer mixed-language content. Cell values and record names automatically detect text direction, making Arabic or Hebrew easier to read alongside English.

AI imaging, performance, and experience optimization

Fixes & Improvements

• More accurate Routine status: Routines show completed promptly after chat generation, reducing service-switching delays during deployment.
• Improved email automation trigger validation: Required-field errors appear only after editing connection fields, reducing premature warnings.
• Improved CLI help: Clarified Skill scope options and environment defaults for easier, more accurate command usage.
• Improved chat tool call display: Tool calls remain visually grouped across steps and subsequent reasoning, improving clarity and traceability.
• Expanded AI image support: Added tailored options and image-to-image capabilities for GPT Image, Gemini, Seedream, Recraft, Muse, Grok Imagine, and Arrow.
• Improved Scrape Data menu: Redesigned icons and layout prevent truncation and enable in-panel configuration for clearer, easier setup.
• Improved Grid loading performance: Reduced unnecessary requests when opening Grids, smoothing status updates during formula, lookup, and rollup calculations.
• Improved form validation: Errors remain in-form instead of appearing unhandled; missing fields consistently show “Required” for quicker resolution.
• Improved AI image generation stability: Enhanced validation, result handling, Arrow SVG output, and preservation of uploaded file types and extensions.
• Email trigger connection changes: “When a new email is received” blocks Gmail/Outlook setup or reauthorization; existing records and IMAP remain.

Automation tasks, chat, and stability upgrades

Feature Updates

• Added Scheduled Routines: Create, schedule, run, pause, and monitor Agent tasks with templates, run history, sidebar navigation, Agent, and CLI tools.

Bug Fixes & Improvements

• Improved Routine scheduling: Refined time zones, daylight saving time, recurrence rules, one-time schedules, and synchronization for reliable execution.
• Improved Routine quota handling: Routines automatically pause after five consecutive quota failures and cancel future runs to prevent waste.
• Improved Routine reliability: Enhanced recovery, deletion cleanup, manual runs, and navigation to activity and run history.
• Enhanced chat management: Web and mobile now support pinning, renaming, archiving, deleting, activity indicators, and manual user sorting.
• Improved chat consistency: Refined message history, tool output, feedback, queued messages, retries, and streaming recovery.
• Improved long-message rendering: Increased streaming response efficiency, especially for mobile reading.
• Improved chat permissions and read status: Unauthorized actions show clear messages; background replies remain unread after switching conversations.
• Improved chat change summaries: Deleted resources now appear unavailable instead of linking to invalid destinations.
• Fixed duplicate image attachments: Resolved duplicate images when pasting content into the upload area.
• Improved attachment editing and recovery: Fixed stuck image replacement and failed-upload recovery, ensuring attachments can be edited, removed, and sent.
• Improved expanded-record interactions: Fixed Single-select dropdown scrolling and prevented accidental closure when pressing Escape.
• Improved field adjustment stability: Deleting or reordering fields while resizing columns no longer crashes the grid or misapplies widths.
• Improved global search results: Search now respects field visibility, excluding hidden-field content from matches and result counts.
• Improved Hosted Connection authorization: Added smoother retries for denied, canceled, blocked, or failed authorization; existing Hosted Gmail authorizations remain supported.
• Improved linked-field cleanup: Deleting linked fields or stored link data now cleans up related links more completely.
• Improved Schema integrity checks: Restored missing tables when valid schema data remains and safely handled partially deleted linked fields.
• Improved performance for high-traffic tables: Reduced duplicate operations and improved concurrent viewing while preserving permissions and fresh-data requirements.
• Improved mobile login and navigation stability: Added system-browser login with PKCE, consent, reusable sessions, shared authentication, and smoother account switching and embedded experiences.
• Improved mobile embedded pages: Simplified table headers, highlighted Settings, and displayed plan, usage, billing, limits, and self-hosted license information read-only.
• Improved calculation issue summaries: Added clearer errors, database context, affected-field counts, occurrence counts, and complete field identifiers.
• Updated Public API reference documentation: Show only endpoints supporting access-token authentication, helping developers find available APIs faster.
• Unified help and plugin documentation links: Standardized links on help.teable.ai while preserving supported deep links.

Security, performance, and stability updates

Feature Updates

• Added secure credential support for Email automations: Email triggers can use authorized credentials, avoiding stored passwords and diagnostic exposure.
• Added Chat resource change summaries: After each conversation, Chat summarizes resources created or modified by the Agent.

Bug Fixes & Improvements

• Improved large-table performance and stability: Reduced lag, timeouts, and failures when editing, inserting, pasting, or reordering records in older views.
• Improved comment count reliability: Comment counts now update more promptly and accurately when records are edited.
• Improved calculation activity refresh stability: Enhanced refreshes during reconnections and high activity while enforcing access permissions.
• Optimized table search: Shows only published, available indexes; correctly falls back when none exist and improves large-table response times.
• Fixed Number field display settings: Bar or ring chart settings are now cleared when switching back to standard number display.
• Fixed Agent operation progress: Progress cards now display correctly when creating an app and performing other actions simultaneously.
• Improved Chat navigation stability: Prevents resources created in other Bases from causing unexpected redirects or persistent processing states.
• Fixed user field mode switching: Grid views load correctly and existing assignments remain when switching between single-select and multi-select.
• Improved App Builder export security: Copies and downloads exclude authorized credentials and connection secrets; owners must provide them separately.

Scraper Expansion and Stability Improvements

Feature Updates

• Expanded Scraper capabilities: Added discovery for X profiles, Facebook content, Amazon sellers, Walmart reviews, and Crunchbase people, with localized labels.

Bug Fixes & Improvements

• Improved long-tail Scraper dataset access: Search shows discovery modes and supports raw dataset IDs with discoverBy.
• Improved Scraper CLI workflow: teable scrape run returns snapshot IDs immediately; use teable scrape status --wait to await completion.
• Improved Scraper safety and billing accuracy: Refined limits and pending-request reuse to prevent duplicate charges during concurrent status queries.
• Improved large-workspace table analysis stability: Reduced database stalls and temporary 503 errors, making analysis more reliable.
• Improved computation task tracking: Ensured consistent batch reruns within stages and clarified serial retries for easier troubleshooting.
• Clarified complexity metric naming: Renamed “Dirty Records” to “Estimated Complexity,” making computation workload easier to assess.
• Improved app and automation credential wording: Renamed secret “Rotate” to “Edit” without changing how updates work.
• Fixed user avatars in credentials and configurations: Owner avatars display correctly, with initials shown when no avatar exists.
• Improved credential usage lists: Shows only additional resources, truncates long names with tooltips, preserves actions, and displays access grant times.
• Improved computed field updates under high concurrency: Enhanced data processing to reduce timeouts and update failures.
• Fixed Lookup and Rollup field inconsistencies: Unified handling of saved and transformed options, cross-Base references, and version changes.
• Improved table loading and connection stability: Reduced unnecessary requests during initial loads, reconnections, and frequent operations for smoother use.
• Fixed computation activity notifications: Prevented subscription errors from triggering unrelated global notifications.
• Fixed User field filters after type changes: Preserves filters and values across selection modes, preventing failures and Grid loading issues.
• Fixed empty-search loading issues: Opening search with Cmd+F no longer loops Grid row loading or resets filters, sorting, and grouping.
• Improved shared view loading reliability: More consistent field and record loading reduces missing or out-of-sync data.

URL recognition, credentials, and search optimization

Feature Updates

• Automatic URL detection: Clickable links work across text fields, including Formula, Rollup, and Long Text, with CJK and touch support.
• Automatic links for AI-created URL fields: New website and URL columns use auto-linking plain text fields; existing configurations remain supported.
• Centralized credential management: Manage reusable personal API keys and third-party connections in Settings › Integrations, with resource permissions and usage tracking.
• Secure credential binding: Apps, automations, and transferred resources retain placeholders; collaborators bind credentials and use aliases for multiple users.
• AI credential confirmation: AI-assisted building requests confirmation before using credentials and masks keys in test output.
• Scraper data access: The chat “+” menu adds a Scraper option for curated datasets, localized prompts, and permission-based access.
• Natural-language Scraper discovery: Find datasets beyond the curated catalog and view required inputs more clearly for faster data collection.

Fixes & Improvements

• Improved all-field search scope: Uses configured indexes, includes newly added fields, and limits unindexed ILIKE searches to 20 eligible fields.
• Improved text search compatibility: Records with inconsistent or unsupported date values are no longer skipped or cause failures.
• Clearer calculation timelines: Improved status labels and task relationships show progress from source updates through processing to completion.
• Fixed idle table calculation status: Idle tables no longer remain on “Checking calculation status” when other tables have queued tasks.
• Improved credential usage display: Summaries now count only hidden resources, with better truncation for long resource names and aliases.
• Improved Scraper reliability: Unavailable posts no longer appear as verification failures, and LinkedIn scraping avoids premature timeouts.
• Improved table query reliability: Applies sensible defaults when query options are omitted, preventing parameter-related errors.
• Faster Table Query Ops overview: Shows only recent data by default, improving page load times and administrative responsiveness.
• Faster calculations and conditional aggregations: Improves high-load processing for quicker data updates and statistical results.
• More stable real-time table activity: Better handles load spikes and delayed calculation activity, reducing connection errors and repeated retries.
• Improved sharing and batch requests: Ignores inaccessible private activity data, preventing one permission issue from failing the entire request.
• Fixed OAuth authorization: App Builder integrations always show authorization prompts, and Google Sheets proceeds beyond selection after authorization.
• Consistent OAuth service support: Aligned App Builder Chat, General Chat, App Builder Agent Computer, and related CLI commands.
• Improved resilience during brief database outages: Reduces intermittent errors across tables, records, comments, sharing, and chat during connection recovery.
• Fixed date-grouped summaries: Group-level values now match visible daily and monthly groups instead of appearing incorrect or empty.
• Fixed filtered comment counts: Corrected legacy single-value date lookup fields while preserving permissions, pagination, collapsed groups, and substring search.

Calculation, Search, and Stability Optimization

Fixes & Improvements

• Improved calculated field stability: Fixed idle tables remaining stuck on “Checking calculation status” after Lookup or linked record updates.
• Improved multi-level calculation accuracy: Fixed missing or delayed downstream formula updates during multi-step Lookup and calculated field recalculation.
• Optimized calculated field performance: Reduced unnecessary calculations, Lookup, and Rollup updates when no records or results changed.
• Improved calculation status messaging: Prevented false warnings, clarified empty tasks, and retained unresolved errors for accurate progress tracking.
• Improved calculation activity reliability: Displayed errors clearly and preserved existing status after check failures or timeouts.
• Prevented redundant invalid tasks: Stopped retries for tasks tied to deleted or missing Bases, improving stability.
• Improved shared document stability: Enhanced error handling for collaboration, loading, and activity replay failures, with clearer permission messages.
• Improved AI settings: Fixed crashes when provider data is missing and allowed administrators to save empty custom provider configurations.
• Improved full-field search performance: Optimized large-table searches across lists, counts, groups, and aggregations, including numbers, rounded values, and short keywords.
• Improved search task responsiveness: Statistical queries now stop when users change or cancel searches, reducing wasted time and resources.
• Improved restricted-user access: Fixed false permission errors for readable calculated fields and improved reliability after permission changes or reconnection.

GPT-6 Is Now in Teable

Query Deduplication and AI Import Optimization

Feature Updates

• Added Lookup field deduplication: Removes duplicate values from multiple linked records for cleaner, more readable results.
• Added AI import recommendations: Prioritizes AI import when creating tables from existing content for faster, more accurate organization.

Bug Fixes & Improvements

• Improved Cuppy context window indicator: Uses a neutral gray progress ring to clarify automatic context compression.
• Improved filter layout: Value inputs use available row width, with better Current User and statistical field filtering.
• Improved public automation Webhook stability: Fixed persistent 429 errors below documented rate limits, ensuring throttling states recover and reset properly.
• Improved AI import reliability: Better handles quota-related rejections and edge cases for more stable, consistent results.
• Optimized large-table search: Improved full-field search across lists, counts, groups, aggregations, shared views, and concurrent queries.
• Optimized statistical query cancellation: Reduces unnecessary processing when search criteria change or queries are canceled, improving responsiveness.
• Improved Rollup conversion stability: Fixed switching from “Unique Array” to “Count All” and validates formats against new result types.
• Improved nested formula stability and performance: Optimized repeated empty checks for Attachment, User, or Lookup fields, reducing computed-field update timeouts.
• Fixed Lookup field filtering: “Contains any → Current User” now correctly filters referenced User fields.

New Pricing: More AI Credits. Save 44% Yearly.

More AI Credits, a deeper yearly discount: yearly billing saves approximately 44% compared with monthly billing, while delivering 50% more AI Credits per dollar than before—making team growth more cost-effective and predictable.

1. What’s Changed

• Pro (Cloud)
  • Monthly: $12 → $24
  • Yearly: $120 → $160 ($10 → $13.33/month)
  • AI Credits: 1,000 → 2,000 per seat/month
• Business (Cloud and Self-hosted)
  • Monthly: $24 → $36
  • Yearly: unchanged at $240 ($20/month)
  • AI Credits (Cloud only): 2,000 → 3,000 per seat/month

2. How Seats and Credits Work

3. Why We’re Making This Change

4. Existing Customers Get More at the Same Price

Automation, attachments, and intelligent collaboration

Feature Updates

• Webhook synchronous responses: Automation triggers support custom responses for Feishu/Lark, Slack, WeCom, and DingTalk URL verification; existing defaults remain unchanged.
• Attachment deletion in previews: Delete files while previewing; automatically switch to adjacent attachments for smoother review and organization.
• Multi-Base Chat agents: Discover and use accessible Bases in the current Space, or specify a target Base per command.

Bug Fixes & Improvements

• Improved computed-field pausing: Allows safe writes during pauses; blocked writes return clear conflict messages and recommended retry times.
• Standardized Secrets Manager terminology: All languages use “Name” for Secret identifiers and “Secret” for values, reducing configuration confusion.
• Stable table and schema updates: Better handles timeouts, missing columns, and rollbacks, reducing failures, misleading messages, and manual recovery.
• Reliable calculations and backfills: Fixed multi-value Lookup, linked-record Rollup, and max/min/and/or failures; timed-out calculations retry automatically.
• Fixed formula field creation: Failed creation no longer makes related tables temporarily unavailable, allowing continued data access and operations.
• Improved Webhook testing and configuration: Tests show received requests; editing some settings preserves authorization and response configurations.
• Linked-record Rollup results: Compressed results show readable titles, preserve duplicates, and match UI/API behavior; integrations using legacy structures may need updates.
• Improved conditional Rollup filters: Preserves nested groups after reopening, works in narrow panels, and aligns Lookup-number results with API filters.
• Attachment preview compatibility: Displays file icons in thumbnails and previews when images fail to render, preventing blank or broken content.
• Chat agent Base detection: Improves discovery for Base-level users and prevents Base context leaking into shared or restored sessions.
• Fixed API Access page: Opening tables from the app sidebar no longer hangs; Advanced now uses the selected table.
• Fixed long text field actions: Referenced fields now open record context menus while preserving native browser menus for selected read-only text.
• Improved Field AI errors: Clarifies cancellation reasons for billing, quotas, authentication, account issues, and image-generation failures.
• Reliable CLI Artifact operations: Fixed operations using personal access or OAuth tokens; permissions and Space/Base restrictions are always enforced.
• Consistent Artifact permissions after moves: Artifacts follow their Base across Spaces and retain applicable access and sharing rules.
• More stable App Builder chat: Improves recovery after interruptions, prevents persistent “Recovering” states, and enhances rich-content display stability and security.
• Improved mobile settings: Refined layouts and navigation for smoother browsing and operation on mobile devices.
• Faster wide-table queries: Requests for limited fields load less unnecessary data, reducing resource usage and accelerating record queries.

Artifact and stability upgrades

Feature Updates

• Cross-session Artifact updates: Agents update existing Artifacts, preserve history, and see names in tabs; Artifact API/CLI dropped legacy chat parameters.
• HEIC attachment thumbnails: Generates PNG previews when supported; otherwise shows placeholders to prevent display issues.
• App Builder publishing guidance: Preview/Live switching is smoother and remembered per app; apps without Live versions receive clear publishing guidance.
• AI field concurrency settings: Space admins can configure concurrency; above-limit values are retained while the effective limit is displayed.

Bug Fixes & Improvements

• Updated API table creation defaults: Tables created without records via API are empty; UI-created tables retain three blank records.
• Faster large-table browsing: Lists and grids load less via cursor pagination; API clients must request totals or use the count endpoint.
• Refined substring search: Prioritizes text and text-returning Formula/Lookup fields; wide-table or unsupported-field results may change.
• Improved record-query performance and stability: High-traffic reads use fewer resources while preserving existing query behavior and improving access efficiency.
• Fixed recent Base ordering and duplicate collaborators: New copies appear first; duplicates merge under the highest available role.
• Improved conditional Rollup configuration: Fixes nested OR and dynamic-match saving/calculation issues and improves narrow-screen controls.
• Improved Rollup validation: APIs enforce UI source-field and aggregation rules with clear errors; invalid legacy configurations require client updates.
• More reliable computed-field updates: Formula, Lookup, Rollup, and summary fields recalculate reliably after linked or dependent data changes.
• Faster conditional Rollup calculations: Multiple Rollups sharing linked records recalculate more efficiently, reducing timeout risks.
• Improved Computed Outbox diagnostics: Reliable task and error links, shared links, and Copy for Agent output accelerate troubleshooting.
• Fixed linked-record selection state: Selections and checkboxes remain synchronized across tab switches, reopened pickers, and asynchronous loading.
• Streamlined Agent Computer browser tasks: AI tasks use built-in browser tools, eliminating downloads and path setup for faster, consistent execution.
• Fixed AI chat external-link dialog layout: Controls no longer overlap, improving readability and usability.
• Improved short-link redirects: Links now redirect directly to canonical targets, reducing intermediate loading and wait time.
• Fixed Auto Number filtering: “Greater than” no longer breaks loading or row counts; saved values display correctly when reopened.
• Secured editable shared views: Record and field changes, including range paste, clear, and delete, are restricted to authorized views.
• Fixed cross-Base Lookup/Rollup configuration: Linked tables remain available when reopened; empty external Base references follow current-Base handling.
• More accurate selected-row summaries: Aggregations reliably honor filters, sorting, search, row ranges, and collapsed groups, with improved Timestamp support.
• Fixed Single Select Rollups: Unique values retain first-seen order; Count includes only unique non-empty values, potentially reducing results.
• Improved linked-field failure resilience: Record creation and schema updates remain stable when imports or deletions create inconsistent references.
• Faster AI-assisted workflows: Reduced redundant refreshes and requests across script editing, publishing, chat, and automations for smoother interactions.
• Fixed record comment permissions: Authorized collaborators can comment; read-only users only view discussions, and shared links hide comment controls.
• Improved comment deletion and image uploads: Deleted comments stay removed; uploads show clearer progress and errors, with sending blocked until completion.
• Improved App Builder workspace recovery: Temporary connection issues preserve workspaces and drafts; corrected version-history alignment prevents publishing failures.
• Improved permission matrix usability: Clearer status icons, tooltips, and better narrow-screen sidebar usability.
• Fixed record archiving and export permissions: Authorized collaborators archive without delete rights; archive exports respect permissions, and views update promptly.
• Unified Skill and Teable CLI instance configuration: Instructions, commands, authentication, settings, and errors now use the current instance address.

Introducing Artifacts: Share Data Reports Anywhere

Turn your Teable data into visual reports with AI, refine them through conversation, and share them with a link.

1. Create Artifacts with AI

2. Keep Refining the Same Artifact

3. Share Artifacts Anywhere