Live service health
All systems operational.
Real-time status and uptime for the services Doe runs, and for the upstream model providers it depends on.
All systems operationalUpdated 9m ago
The systems we run.
Backend
Backend Servers
465msOperational90 days agoToday90-day uptime99.92%Frontend
Frontend Servers
724msOperational90 days agoToday90-day uptime99.87%
The providers we depend on.
Upstream model providers, monitored through their official status feeds. Trouble here doesn't count as a Doe outage, but it can slow work that routes through them.
OpenAI API
Upstream provider (status.openai.com)
Operational90 days agoToday90-day uptime97.13%Anthropic API
Upstream provider (status.anthropic.com)
Operational90 days agoToday90-day uptime98.71%AWS Bedrock
Upstream provider (AWS Health)
Operational90 days agoToday90-day uptime98.42%
Backend latency, last 24 hours.
- Average
- 334ms
- Min
- 49ms
- Max
- 1835ms
What shipped.
Nov 14, 09:52 AM
Update
Implemented advanced coding capabilities
- chore: bump frontend version to v1.0.498 [skip ci] - [eng-2620] lead execute_code bash rows with the model's description (#2508) - chore: bump backend version to v1.9.570 [skip ci] - chore: bump frontend version to v1.0.497 [skip ci] - [eng-2594] route fable mode bedrock requests to fable 5.2 (#2473) - chore: bump frontend version to v1.0.496 [skip ci] - chore: bump backend version to v1.9.569 [skip ci] - [eng-2605] do not push a pause prompt to the slack sink twice (#2507) - chore: bump frontend version to v1.0.495 [skip ci] - [eng-2619] keep inline file and component markers inside Markdown table cells without splitting rows (#2506) - chore: bump backend version to v1.9.568 [skip ci] - chore: bump frontend version to v1.0.494 [skip ci] - [eng-2610] lead shell activity rows with a model-written description and collapse the command (#2487) - chore: bump backend version to v1.9.567 [skip ci] - [eng-2618] fix flaky desktop viewer upgrade test: tamper a signature character that cannot decode back to the original (#2505) - [eng-2614] let import_repo pin a commit sha and report git's reason when a clone fails (#2501) - [eng-2612] bound the coding shell writer wait and let an unlaunched start settle so a failed checkpoint start cannot wedge a workspace (#2499) - [eng-2616] nudge and auto-checkpoint coding workspaces, and restart the jobs a save stops (#2503) - chore: bump frontend version to v1.0.493 [skip ci] - chore: bump backend version to v1.9.566 [skip ci] - chore: bump frontend version to v1.0.492 [skip ci] - [eng-2604] shorten user-message previews, clarify code controls, and smooth expansion with reduced-motion support (#2481) - [eng-2608] keep the gate on an adopted pause row and attach the workspace before touching a resumed sandbox (#2485) - chore: bump backend version to v1.9.565 [skip ci] - chore: bump frontend version to v1.0.491 [skip ci] - [eng-1937] fix false-offline detection, hydrate scoped desktop caches, recover offline previews (#2460) - chore: bump frontend version to v1.0.490 [skip ci] - [eng-2611] explain silent recordings instead of failing transcription, and make the transcript panel resizable and collapsible (#2488) - chore: bump frontend version to v1.0.489 [skip ci] - [eng-2609] fix missing sender profile pictures in creator-only Slack threads, preserve group attribution (#2486) - chore: bump backend version to v1.9.564 [skip ci] - [eng-2607] ask for workspace_transfer approval only when it overwrites and the user has no write grant (#2484) - chore: bump frontend version to v1.0.488 [skip ci] - [eng-2599] add library, side session, session info, insights and canvas to the preview pane add menu (#2474) - [eng-2598] serve pull request cards from a stored snapshot and prefetch them when a chat opens (#2475) - chore: bump backend version to v1.9.563 [skip ci] - [eng-2605] post slack pause suggestions after the pause message, not before it (#2482) - [eng-2600] write specific slack thinking-step labels with an llm instead of generic tool-name fallbacks (#2476) - [eng-2602] render chat-file images as image blocks in slack replies (#2479) - [eng-2603] coding agent announces its local-stack plan and runs applications with scratch services efficiently (#2480) - chore: bump backend version to v1.9.562 [skip ci] - record judgment conversation observations on llm spans and fix two trace health failures (#2478) - chore: bump backend version to v1.9.561 [skip ci] - chore: bump frontend version to v1.0.487 [skip ci] - [eng-2601] authenticate the computer view's desktop socket with the session token and surface refused connections (#2477) - chore: bump frontend version to v1.0.486 [skip ci] - chore: bump backend version to v1.9.560 [skip ci] - chore: bump frontend version to v1.0.485 [skip ci] - chore: bump frontend version to v1.0.484 [skip ci] - [eng-2595] patch critical dependencies and enforce template, proxy and git security regressions (#2469) - chore: bump backend version to v1.9.559 [skip ci] - [eng-2596] queue github pull request activity behind a running turn instead of injecting it (#2471) - [eng-2597] hide doe's bookkeeping from slack thinking steps and never show a step as failed (#2472) - chore: bump frontend version to v1.0.483 [skip ci] - [eng-2595] fix high-severity dependencies, keep tsx imports executable and preserve production dependency trees (#2470) - chore: bump backend version to v1.9.558 [skip ci] - [eng-2593] hide doe's follow-up turns from the user's web message queue (#2468) - [eng-2591] record abort provenance and name the opened pull requests when a turn dies (#2467) - [eng-2589] release repository workspace leases when the turn ends or is stopped (#2465) - [eng-2590] end the judge loop on an unreachable verdict and keep judge notes out of the user's reply (#2466) - chore: bump frontend version to v1.0.482 [skip ci] - chore: bump backend version to v1.9.557 [skip ci] - chore: bump frontend version to v1.0.481 [skip ci] - [eng-1562] ingest forwarded Slack images, retain scoped attachments during active turns (#2459) - chore: bump backend version to v1.9.556 [skip ci] - [eng-2588] post doe's follow-up turns to slack as doe, not under the user's name (#2463) - chore: bump frontend version to v1.0.480 [skip ci] - [eng-2586] show preloaded guidance and skills in the activity feed (#2461) - [eng-2587] memory follow-ups: json errors on knowledge routes, confidence on reinforcement, share learned memories (#2462) - [eng-2592] restore the lockfile entries the backend version bump dropped and pin npm for the bump (#2464) - chore: bump backend version to v1.9.555 [skip ci] - [eng-2585] explain to the user why a repository secret cannot run in execute_code instead of advertising it (#2458) - chore: bump frontend version to v1.0.479 [skip ci] - chore: bump backend version to v1.9.554 [skip ci] - chore: bump frontend version to v1.0.478 [skip ci] - [eng-2581] one in-house connector registry, github in-house only, github tool outputs kept for the judge (#2454) - [eng-2579] show the repositories settings tab to every organization and let admins turn the coding agent on (#2451) - [eng-2582] fail a coding preparation whose exited command never seals, and version the dev media bucket (#2455) - chore: bump frontend version to v1.0.477 [skip ci] - [eng-2580] share memories across the organization (#2453) - chore: bump frontend version to v1.0.476 [skip ci] - chore: bump backend version to v1.9.553 [skip ci] - [eng-2577] keep a verified publish, check a missing-deliverable disclosure against the record, say when a browser tool is gated (#2447) - chore: bump frontend version to v1.0.475 [skip ci] - [eng-2570] land the coding agent tools, sandbox runtime, github app layer and executor wiring behind the organization flag (#2445) - [eng-2568] land the coding guest programs, their generator and the program cache with a ci staleness check (#2441) - chore: bump backend version to v1.9.552 [skip ci] - [eng-2571] record which app made each installation row, and act only on this deployment's rows - [eng-2571] name the dev app's handle, command and check as its slug derives them - [eng-2571] record the dev app as created: doe-labs-dev, installable on the Doe-Labs account only - [eng-2571] say that user authorization during installation stays off - [eng-2571] keep production's callback url off the dev task - [eng-2571] run each deployment as its own github app, and derive every github-facing name from that app's slug - [eng-2567] keep the migration helper out of the deploy's pathspec, refuse the downs that would strip queued Build cleanups or standalone coding rows, and leave the installation uniqueness change to the slice that changes the writer - [eng-2567] land the coding agent schema as six additive migrations whose downs refuse to lose rows - [eng-2566] add the per-organization coding agent flag and the helper every coding slice checks - chore: bump backend version to v1.9.551 [skip ci] - chore: bump desktop version to v0.1.50 [skip ci] - chore: bump frontend version to v1.0.474 [skip ci]
Nov 4, 05:43 AM
Update
Minor update
- Redesigned the Sessions page with folder grouping, pinned and archived sections, drag-and-drop organization, and customizable layouts - Added native Vanta integration with OAuth - Improved agent reliability with safer approvals, more accurate billing, and better recovery for long-running tasks - Strengthened security with stricter file permissions, safer tool execution, and improved data isolation - Improved search performance, session controls, and tool result handling
Nov 3, 03:05 AM
Bug Fix
Minor fix
- Improved approval clarity with accurate email draft labels and safer Google Sheets write controls - Proactive tasks now wait until onboarding is complete before running automatically - Fixed message clipping on smaller mobile screens - Improved agent responses by preventing placeholder text from appearing in final answers
Oct 30, 06:15 PM
Update
Minor update
- Improved integration reliability by preserving native MCP tools across mixed tool catalogs - Fixed document persistence test timeouts for more reliable deployments
Oct 30, 09:29 AM
Update
Minor update
- iMessage now supports tool approvals and replies directly within conversations - Agents can download Gmail attachments into the Library and handle email verification more securely - Proactive tasks can now be disabled through natural language - Improved Asana reliability and iMessage contact branding
Oct 30, 04:56 AM
Update
Minor update
- Added Affinity integration with OAuth and API key support - Improved sidebar loading with smoother pagination and fewer session list resets - More reliable ZIP uploads with stronger file validation and security checks
Oct 29, 05:37 PM
Update
Minor update
- Added Harmonic integration with OAuth and API key authentication options
Oct 28, 09:21 AM
Bug Fix
Minor fix
- Improved sidebar loading reliability by excluding system folders from session fetching
Oct 24, 03:04 AM
Update
Minor update
- Folder sorting preferences now sync across devices, with more accurate recent activity ordering - Improved session organization with smoother keyboard shortcuts and folder menus - Fixed mobile menus, file downloads, and access to shared files across sessions - Restored organization-wide sharing controls for native integrations
Oct 21, 04:42 AM
Update
Minor update
- Session previews now automatically advance to the next conversation after archiving - Pending tool approvals can now be accepted by pressing Enter
Oct 20, 10:21 AM
Update
Minor update
- Improved Slack integration with native stop controls and two-way session renaming - Added a keyboard shortcut for moving sessions between folders - Refined folder menus and expanded subagent display names
Oct 16, 05:51 AM
Update
Minor update
- Added interactive session previews and side chats that reference the main conversation - Sidebar can now be resized, with synced preferences, back/forward navigation, and draft indicators - Standardized navigation across the app with consistent top bars and improved keyboard shortcuts
Oct 10, 01:25 AM
Update
Implemented sessions page
- Added a dedicated Sessions page with filters, previews, and bulk actions - Improved sidebar organization with collapsible sections - More transparent proactive tasks with visible data sources and manageable tool approvals - Fixed email formatting issues with plain-text messages
Oct 8, 03:00 AM
Update
Minor update
- Proactive tasks now handle safe actions automatically and learn from previously approved tools - Improved website publishing reliability for larger files - More efficient execution environments with automatic sleep and flexible storage - Fixed session interruptions and unread states during folder reorganization
Oct 7, 01:23 AM
Bug Fix
Minor fix
- Agents can now retrieve, edit, list, and delete saved Gmail drafts - LinkedIn connections now use the native OAuth flow throughout the app - Improved Cloudflare and Granola connection reliability - Fixed response validation incorrectly blocking valid replies
Sep 29, 07:27 AM
Improvement
Minor improvement
- Improved code execution for large outputs and background processes
Sep 29, 04:18 AM
Improvement
Minor improvement
- Faster agent responses by skipping unnecessary synthesis when a complete answer is already available - Improved document exports with built-in validation and quality checks - More reliable Slack workflows during API rate limits
Sep 25, 06:24 AM
Bug Fix
Minor fix
- Updated LinkedIn Ads to the latest API version for continued integration reliability
Sep 22, 04:45 AM
Bug Fix
Minor fix
- Fixed repeated sign-in prompts after authorizing the hosted CLI - Improved agent email formatting for long HTML emails
Sep 21, 09:20 AM
Bug Fix
Minor fix
- Improved GPT-6 Astra reliability with better retry and provider failover handling
The last 90 days.
November 13, 2026
- Resolved
AWS Bedrock (global): operational issue
[10:52 AM PST] We are investigating Claude model issues for Bedrock. During this time, you may experience failures or increased response times when invoking these models, including requests made through cross-region inference. [11:00 AM PST] We are experiencing increased error rates for some Claude models on Amazon Bedrock. During this time, you may experience failures or increased response times when invoking these models, including requests made through cross-region inference. We are actively working on identifying the root cause and mitigating the issue. We will provide another update by 12:30 PM PST, or sooner if more information becomes available. [12:39 PM PST] We are seeing recovery for increased error rates for the Claude Opus 4.8 model and we continue to work on identifying the root cause and resolving the issue for the Claude Opus 5 and Claude Opus 5.5 models. Customers may experience failures or increased response times when invoking Claude Opus 5 and Claude Opus 5.5. This also impacts requests made through cross-region inference for these models. In the meantime, customers may be able to use Claude Opus 4.8 model. We will provide another update by 2:00 PM PST, or sooner if more information becomes available. [2:01 PM PST] We continue to work on resolving the increased error rates issue for the Claude Opus 5 and Claude Opus 5.5 models. We have identified the root cause and are actively working on mitigating the issue. Our mitigation efforts are underway and we are making progress. In the meantime, customers may be able to use Claude Opus 4.8 model. We will provide another update by 4:30 PM PST, or sooner if more information becomes available. [4:30 PM PST] Starting at 4:30 AM PST, we began experiencing increased error rates for some Claude models. Customers may have experienced failures or increased response times when invoking Claude Opus 4.8, Claude Opus 5 and Claude Opus 5.5 models, including requests made through cross-region inference (CRIS). No other models were impacted. Our engineers were engaged and identified the cause and implemented mitigation steps to achieve full recovery for Global CRIS customers. For Opus 4.8 the impact was between 9:00 AM and 11:25 AM, the Opus 5 impact was between 4:30 AM and 2:10 PM, and the Opus 5.5 impact was between 8:30 AM and 3:10 PM. We recommend Global CRIS customers retry any failed requests during these impact windows. However we continue to see Increased error rates for Claude Opus 5.5 requests using the in-region (AP-SOUTHEAST-4) bedrock-mantle endpoint or the AU Geo CRIS inference profile. Where possible, we recommend these customers use the US Geo CRIS inference profile instead until recovery is complete. We will provide another update by 7:30 PM PST, or sooner if more information becomes available.
minorStarted 06:57 PMResolved 02:15 AM7h 18m
- Resolved
AWS Bedrock (global): operational issue
We are investigating Claude model issues for Bedrock. During this time, you may experience failures or increased response times when invoking these models, including requests made through cross-region inference.
minorStarted 06:57 PMResolved 02:15 AM7h 18m
- Resolved
AWS Bedrock (us-east-1): model issue
[9:56 AM PST] We are investigating Claude model issues for Bedrock in the US-EAST-1 Region. During this time, you may experience failures or increased response times when invoking these models, including requests made through cross-region inference. [10:53 AM PST] We are experiencing increased error rates for some Claude models on Amazon Bedrock. During this time, you may experience failures or increased response times when invoking these models, including requests made through cross-region inference. We are actively working on identifying the root cause and mitigating the issue. We will provide another update by 12:30 PM PST, or sooner if more information becomes available. [12:39 PM PST] We are seeing recovery for increased error rates for the Claude Opus 4.8 model and we continue to work on identifying the root cause and resolving the issue for the Claude Opus 5 and Claude Opus 5.5 models. Customers may experience failures or increased response times when invoking Claude Opus 5 and Claude Opus 5.5. This also impacts requests made through cross-region inference for these models. In the meantime, customers may be able to use Claude Opus 4.8 model. We will provide another update by 2:00 PM PST, or sooner if more information becomes available. [2:00 PM PST] We continue to work on resolving the increased error rates issue for the Claude Opus 5 and Claude Opus 5.5 models. We have identified the root cause and are actively working on mitigating the issue. Our mitigation efforts are underway and we are making progress. In the meantime, customers may be able to use Claude Opus 4.8 model. We will provide another update by 4:30 PM PST, or sooner if more information becomes available. [4:32 PM PST] Starting at 4:30 AM PST, we began experiencing increased error rates for some Claude models in the US-EAST-1 Region. Customers may have experienced failures or increased response times when invoking Claude Opus 4.8, Claude Opus 5 and Claude Opus 5.5 models, including requests made through cross-region inference (CRIS). No other models were impacted. Our engineers were engaged and identified the cause and implemented mitigation steps to achieve full recovery for Global CRIS customers. For Opus 4.8 the impact was between 9:00 AM and 11:25 AM, the Opus 5 impact was between 4:30 AM and 2:10 PM, and the Opus 5.5 impact was between 8:30 AM and 3:10 PM. We recommend Global CRIS customers retry any failed requests during these impact windows. However we continue to see Increased error rates for Claude Opus 5.5 requests using the in-region (AP-SOUTHEAST-4) bedrock-mantle endpoint or the AU Geo CRIS inference profile. Where possible, we recommend these customers use the US Geo CRIS inference profile instead until recovery is complete. We will provide another update by 7:30 PM PST, or sooner if more information becomes available.
minorStarted 05:58 PMResolved 02:18 AM8h 19m
- Resolved
AWS Bedrock (us-east-1): model issue
[9:56 AM PST] We are investigating Claude model issues for Bedrock in the US-EAST-1 Region. During this time, you may experience failures or increased response times when invoking these models, including requests made through cross-region inference.
minorStarted 05:58 PMResolved 02:18 AM8h 19m
November 10, 2026
- Resolved
Backend is experiencing issues
Connection timeout or failure
majorStarted 09:10 AMResolved 09:15 AM5m
October 31, 2026
- Resolved
Frontend is experiencing issues
Connection timeout or failure
majorStarted 06:32 PMResolved 06:37 PM5m
October 30, 2026
- Resolved
AWS Bedrock (us-east-1): model issue
[9:18 AM PDT] We are investigating Bedrock model issues for Anthropic Claude Opus 5.5 in the US-EAST-1 Region. [9:47 AM PDT] Starting at 8:40 AM PDT, we are experiencing Bedrock model issues for Anthropic Claude Opus 5.5 in the US-EAST-1 Region. This will also impact requests routed to US-EAST-1 through cross-Region inference (CRIS). During this time, you may experience elevated latency, timeouts, or intermittent errors when invoking these models, including streaming responses that end early. We have identified the root cause and are actively working on mitigating the issue. We will provide another update by 11:30 AM PDT or sooner when we have more information to share.
minorStarted 05:28 PMResolved 05:46 PM17m
- Resolved
AWS Bedrock (us-east-1): model issue
[9:18 AM PDT] We are investigating Bedrock model issues for Anthropic Claude Opus 5.5 in the US-EAST-1 Region. [9:47 AM PDT] Starting at 8:40 AM PDT, we are experiencing Bedrock model issues for Anthropic Claude Opus 5.5 in the US-EAST-1 Region. This will also impact requests routed to US-EAST-1 through cross-Region inference (CRIS). During this time, you may experience elevated latency, timeouts, or intermittent errors when invoking these models, including streaming responses that end early. We have identified the root cause and are actively working on mitigating the issue. We will provide another update by 11:30 AM PDT or sooner when we have more information to share.
minorStarted 05:28 PMResolved 05:46 PM17m
October 29, 2026
- Resolved
Frontend is experiencing issues
Connection timeout or failure
majorStarted 10:10 PMResolved 10:16 PM5m
October 28, 2026
- Resolved
AWS Bedrock (us-west-2): model issue
We are investigating Bedrock model issues in the US-WEST-2 Region.
minorStarted 07:22 PMResolved 07:39 PM17m
October 23, 2026
- Resolved
Anthropic API is experiencing issues
API components: major_outage
majorStarted 11:39 PMResolved 12:20 AM40m
October 14, 2026
- Resolved
AWS Bedrock (us-east-1): operational issue
[October 14 2:28 AM PDT] Starting at October 13 9:30 PM PDT, we are experiencing increased error rates for Claude Opus 5.5 in the US-EAST-1 Region. Customers may observe 5xx errors. Our engineers were immediately engaged, found the root cause and are actively working on implementing mitigation strategies. As a workaround, customers can retry any failed requests with exponential backoff. We recommend customers try switching to other Claude models if it is possible. We will provide a further update by October 14 4:00 AM PDT or sooner as more information becomes available. [October 14 3:54 AM PDT] Our engineers have applied a mitigation for the increased error rates affecting Claude Opus 5.5 in the US-EAST-1 Region and the service is in a stable state. A subset of requests may continue to experience elevated error rates while we complete our mitigation work. As a workaround, customers can retry failed requests with exponential backoff or switch to other Claude models if possible. We will provide a further update by October 14 10:00 AM PDT or sooner as more information becomes available.
minorStarted 09:29 AMResolved 12:36 PM3h 6m
October 13, 2026
- Resolved
Backend is experiencing issues
Connection timeout or failure
majorStarted 09:10 AMResolved 09:27 AM17m
- Resolved
Backend is experiencing issues
Connection timeout or failure
majorStarted 12:30 AMResolved 12:35 AM5m
October 6, 2026
- Resolved
AWS Bedrock (us-east-1): operational issue
[7:55 AM PDT] Beginning at 7:15 AM PDT, we are experiencing increased error rates for the Claude Opus 5.5 model in the US-EAST-1 Region. Some requests to Claude Opus 5.5 are returning HTTP 503 service unavailable errors, and some streaming responses are ending with an error before completing. Requests from other Regions that are served in the US-EAST-1 Region through cross-region inference may also be affected. We are actively working to mitigate the issue. Customers are advised to retry any failed requests using exponential backoff. We will provide another update by 9:30 AM PDT or as soon as more information becomes available. [9:37 AM PDT] We continue to investigate increased error rates for the Claude Opus 5.5 model in the US-EAST-1 Region. Customers may see some requests to Claude Opus 5.5 returning HTTP 503 service unavailable errors, and some streaming responses ending with an error before completing. Requests from other Regions that are served in the US-EAST-1 Region through cross-region inference may also be affected. We are actively working to mitigate the remaining issue. Customers are advised to retry any failed requests using exponential backoff. We will provide another update by 11:00 AM PDT or sooner when more information becomes available. [10:58 AM PDT] We continue to work towards resolving the issue of increased error rates for the Claude Opus 5.5 model in the US-EAST-1 Region. We have identified the root cause and we are actively working on mitigating the issue. Customers are advised to retry any failed requests using exponential backoff. We will provide another update by 1:00 PM PDT or sooner when more information becomes available.
minorStarted 02:57 PMResolved 08:09 PM5h 12m
October 2, 2026
- Resolved
AWS Bedrock (us-east-1): operational issue
[4:22 AM PDT] Starting at 3:06 AM PDT, we are investigating increased error rates for the Claude Opus 5.5 model in the US-EAST-1 Region. During this time, some requests to Claude Opus 5.5 may return HTTP 503/529 errors, and some requests may take longer than normal to complete. Requests from other AWS Regions served through cross-region inference in US-EAST-1 may also be affected. Our engineering team is actively working to mitigate the issue. Customers are advised to retry failed requests using exponential backoff. We will provide another update by 6:00 AM PDT, or sooner if we have additional information to share.
minorStarted 11:23 AMResolved 11:45 AM21m
September 24, 2026
- Resolved
Backend is experiencing issues
Connection timeout or failure
majorStarted 04:38 PMResolved 04:42 PM3m
September 13, 2026
- Resolved
Backend is experiencing issues
Connection timeout or failure
majorStarted 02:24 AMResolved 02:29 AM5m
September 12, 2026
- Resolved
Frontend is experiencing issues
Connection timeout or failure
majorStarted 07:02 AMResolved 07:07 AM4m
September 10, 2026
- Resolved
AWS Bedrock (us-west-2): operational issue
We are investigating availability issues affecting GPT-6 Astra model in the US-WEST-2 Region.
minorStarted 07:55 PMResolved 08:41 PM46m
September 9, 2026
- Resolved
Frontend is experiencing issues
Connection timeout or failure
majorStarted 07:52 AMResolved 07:57 AM5m
September 7, 2026
- Resolved
Backend is experiencing issues
Connection timeout or failure
majorStarted 11:32 PMResolved 11:37 PM5m
September 4, 2026
- Resolved
AWS Bedrock (us-east-1): anthropic 503 fault
We are investigating increased 503 service unavailable errors for some Claude models in the us-east-1 Region. [3:39 PM PDT] We are experiencing intermittent periods of increased 503 service unavailable errors and throttling for the Haiku 4.5 model. We are working to identify root cause while we mitigate the issue. We will provide you with another update by 6:00 PM PDT or sooner if new information becomes available. [6:14 PM PDT] We continue to work to resolve intermittent 503 Service Unavailable errors and throttling issues for the Haiku 4.5 model in the US-EAST-1 Region. We have identified the root cause and are actively applying mitigation steps. We recommend customers retry any failed requests. We will provide an update by 9:00 PM PDT or sooner if we have further information to share.
minorStarted 10:09 PMResolved 02:01 AM3h 51m
- Resolved
AWS Bedrock (us-east-1): anthropic 503 fault
We are investigating increased 503 service unavailable errors for some Claude models in the us-east-1 Region.
minorStarted 02:03 PMResolved 02:24 PM20m
- Resolved
AWS Bedrock (us-west-2): anthropic 503 fault
We are investigating increased 503 service unavailable errors for some Claude models in the us-west-2 Region.
minorStarted 02:02 PMResolved 02:24 PM21m
- Resolved
AWS Bedrock (us-west-2): anthropic 503 fault
We are investigating increased 503 service unavailable errors for some Claude models in the us-west-2 Region.
minorStarted 02:02 PMResolved 02:24 PM21m
August 27, 2026
- Resolved
AWS Bedrock (us-west-2): anthropic 503 fault
[11:37 AM PDT] We are investigating increased 503 service unavailable errors for some Claude models in the US-WEST-2 Region. [12:03 PM PDT] Starting 10:58 AM PDT, we are experiencing increased 503 service unavailable errors for the Anthropic Claude Haiku 4.5 model in the US-WEST-2 Regions. We are actively engaged to mitigate this issue while identifying the root cause in parallel. We will provide another update by 1:00 PM PDT or as soon as new information becomes available.
minorStarted 07:12 PMResolved 07:49 PM36m
- Resolved
AWS Bedrock (us-west-2): anthropic 503 fault
[11:37 AM PDT] We are investigating increased 503 service unavailable errors for some Claude models in the US-WEST-2 Region. [12:03 PM PDT] Starting 10:58 AM PDT, we are experiencing increased 503 service unavailable errors for the Anthropic Claude Haiku 4.5 model in the US-WEST-2 Regions. We are actively engaged to mitigate this issue while identifying the root cause in parallel. We will provide another update by 1:00 PM PDT or as soon as new information becomes available.
minorStarted 07:12 PMResolved 07:49 PM36m
August 24, 2026
- Resolved
Backend is experiencing issues
Connection timeout or failure
majorStarted 04:56 AMResolved 05:01 AM5m