https://radar.trulysovereign.ai Items whose recommended action is migrate. en Thu, 24 Sep 2026 04:30:30 GMT https://radar.trulysovereign.ai/items/9d3c2cca-9339-4d39-aaee-5059fb7eb195 9d3c2cca-9339-4d39-aaee-5059fb7eb195 Wed, 23 Sep 2026 21:12:12 GMT migrate Claude Platform release notes Claude Opus 5.5 (claude-opus-5-5) launched with 1M token context, 128k max output, $4/$20 per MTok (down from Opus 5's $5/$25). On this model, thinking: {"type": "disabled"} and thinking: {"type": "enabled"} now return 400 errors; tool_choice types "any" and "tool" also return 400. On Claude API and Google Cloud, computer_20251124 tool returns 400 and requires computer_toolset_20260801 instead. Amazon Bedrock still accepts computer_20251124. Why it matters: Existing code that sets thinking parameters or uses tool_choice "any"/"tool" will break with 400 errors if you switch to Opus 5.5. Code using computer_20251124 on Claude API or Google Cloud will also break. If you have hardcoded model names or auto-upgrade logic, you may hit these errors without intending to migrate. What to do: Grep your codebase for "thinking": {"type" and tool_choice references. If found, pin to claude-opus-5 in model selection logic until you can remove those parameters and migrate computer use toolsets per the migration guide. If you do not use these features, verify your model selection does not auto-upgrade, then test Opus 5.5 in staging to confirm 20% cost savings apply to your workload. https://radar.trulysovereign.ai/items/f03f2087-0018-42ea-940a-06772654a4a4 f03f2087-0018-42ea-940a-06772654a4a4 Tue, 22 Sep 2026 16:28:18 GMT migrate OpenRouter model catalogue The Z.ai GLM 5.3 batch model reduced max_completion_tokens from 943,718 to 131,072 tokens (86% reduction). Prompt pricing increased from $0.0000007 to $0.00000072 per token (2.9% increase). Completion pricing increased from $0.0000022 to $0.0000024 per token (9.1% increase). Context length remains 1,048,576 tokens. Why it matters: Any batch job generating outputs longer than 131K tokens will now fail or truncate. If you were relying on the previous 943K output capacity for long-form generation, summarization of large documents, or multi-document synthesis, those workflows will break. The pricing increase affects cost forecasts for high-volume batch processing. What to do: Grep your codebase for 'glm-5.3:batch' and check max_tokens or completion length assumptions. If any job expects outputs over 131K tokens, either chunk the work into multiple calls or switch models. Recalculate batch processing costs using the new rates and update budgets if this model handles significant volume. https://radar.trulysovereign.ai/items/da960a09-8d62-4cd9-aa80-fc01b818f5fc da960a09-8d62-4cd9-aa80-fc01b818f5fc Mon, 21 Sep 2026 23:43:23 GMT migrate Azure Foundry model retirement schedule The retirement date for gpt-4.1-nano moved from 2026-10-14 to 2027-04-14, a six-month extension. The model entered Legacy status on 2025-04-14. Why it matters: If you are running gpt-4.1-nano in production, you now have until mid-April 2027 to complete migration instead of mid-October 2026. This changes the timeline for any scheduled replacement work. What to do: Check your deployment manifests and infrastructure-as-code for references to gpt-4.1-nano. If found, update your migration plan to reflect the new 2027-04-14 deadline and communicate the revised timeline to stakeholders. https://radar.trulysovereign.ai/items/1025bd98-ec2e-4bba-a135-160375cd0aa8 1025bd98-ec2e-4bba-a135-160375cd0aa8 Mon, 21 Sep 2026 23:42:57 GMT migrate Azure Foundry model retirement schedule gpt-4.1-nano retirement moved from 2027-04-14 to 2026-10-14. gpt-image-1.5 retirement moved from 2027-06-16 to 2026-12-16. Both dates accelerated by six months. Why it matters: Any deployments using these models will stop working on the new dates. Six months less migration runway means work scheduled for Q2 2027 must now complete by Q4 2026. What to do: Audit Azure deployments for gpt-4.1-nano and gpt-image-1.5. If found, schedule migration to successor models before 2026-10-14 and 2026-12-16 respectively. https://radar.trulysovereign.ai/items/e0890059-8377-432d-a8bb-375aa1351c22 e0890059-8377-432d-a8bb-375aa1351c22 Sat, 12 Sep 2026 10:13:50 GMT migrate OpenAI deprecations OpenAI announced on September 11, 2026 that gpt-5.4-cyber is deprecated and will be removed from the API on October 1, 2026. The replacement model is gpt-5.6-cyber. Why it matters: Any code calling gpt-5.4-cyber will fail after October 1, 2026. You have 19 days to complete the migration. What to do: Grep your codebase for 'gpt-5.4-cyber', update all references to 'gpt-5.6-cyber', test the replacement model's output against your use cases, and deploy before October 1. https://radar.trulysovereign.ai/items/23bc0de3-09e4-4b40-8fef-f1254dbbec6d 23bc0de3-09e4-4b40-8fef-f1254dbbec6d Sat, 12 Sep 2026 03:31:37 GMT migrate Azure Foundry model retirement schedule o1-pro, o3, o3-deep-research, and o3-pro moved from GA to Deprecated status. All o1/o3 models now retire 2026-11-19 (previously scattered between 2026-10-01 and 2026-12-26). o3-mini and o4-mini successor changed from o4-mini to gpt-5.6-terra. Fine-tuning retirement for o4-mini shifted from 2027-04-16 to 2027-05-19. Why it matters: If you are running o1-pro, o3, o3-deep-research, or o3-pro in production, they are now deprecated with 9 weeks until retirement. The consolidated sunset date means you cannot stagger migrations. The successor model change means existing migration plans targeting o4-mini must be rewritten to target gpt-5.6-terra or gpt-5.6-sol. What to do: Grep your codebase for model identifiers o1-pro, o3, o3-deep-research, o3-pro, o3-mini, o4-mini. If any are in use, schedule migration to gpt-5.6-sol (for o1/o3 reasoning models) or gpt-5.6-terra (for o3-mini/o4-mini) before 2026-11-19. If you have fine-tuned o4-mini models, note the new training retirement date of 2027-05-19. https://radar.trulysovereign.ai/items/c5855813-268c-4213-8f94-7cf3f38ef45c c5855813-268c-4213-8f94-7cf3f38ef45c Thu, 03 Sep 2026 03:20:06 GMT migrate Azure Foundry model retirement schedule Azure Foundry added a retirement date of 2027-12-05 for the claude-fable-5-1 model. The model is currently in Preview status. Why it matters: Any code or configuration using claude-fable-5-1 by name will stop working after that date. Preview models typically have shorter lifespans than GA releases, and 15 months is the notice window. What to do: Grep your codebase and infrastructure-as-code for 'claude-fable-5-1' references. If found, add a calendar reminder for Q3 2027 to migrate to whatever replaces it, or pin to claude-fable-5 if that model remains supported. https://radar.trulysovereign.ai/items/0ee7793d-d108-4471-9679-5872fdd2c0ee 0ee7793d-d108-4471-9679-5872fdd2c0ee Thu, 03 Sep 2026 03:19:57 GMT migrate Claude Platform release notes The Admin API (user management and spend limits), Analytics API, and Compliance API now require the anthropic-version header on every request, matching the existing requirement across the rest of the Claude API. Documentation updated September 1, 2026. Why it matters: Requests to these three Enterprise endpoints will fail without the header. If your integration calls these endpoints and does not currently send anthropic-version, those calls are now broken. What to do: Grep your codebase for calls to Admin, Analytics, and Compliance API endpoints. Add the anthropic-version header to each request if missing. Test against production to confirm the calls succeed. https://radar.trulysovereign.ai/items/cf2722ee-e40f-454a-bc92-5775238664a9 cf2722ee-e40f-454a-bc92-5775238664a9 Tue, 01 Sep 2026 20:40:33 GMT migrate Azure Foundry model retirement schedule claude-opus-4-1 marked Retired with effective date 2026-08-05. claude-opus-5 added to supported models list. Two new preview models added: claude-fable-5-1 and claude-mythos-5-1. Why it matters: Any code or configuration explicitly requesting claude-opus-4-1 will fail after August 5, 2026. If you are using version pinning or explicit model identifiers rather than latest-stable aliases, calls will return errors once the retirement date passes. What to do: Grep your codebase and infrastructure-as-code for 'claude-opus-4-1' and replace with 'claude-opus-5'. Test the replacement in staging before August 5, 2026. https://radar.trulysovereign.ai/items/ba917d39-270a-4508-8b5e-ddc025e3fd7c ba917d39-270a-4508-8b5e-ddc025e3fd7c Tue, 01 Sep 2026 20:40:24 GMT migrate Claude Platform release notes Claude Fable 5.1 and Mythos 5.1 launched September 1, 2026. tool_choice types 'any' and 'tool' now return 400 errors; only 'auto' and 'none' work. Cache read pricing dropped from 0.1x to 0.025x base price ($0.25/MTok). Both models enforce 30-day minimum data retention with no zero-retention option unless Anthropic authorizes it. New accounts created August 31, 2026 or later get 400 errors when replaying thinking blocks after system prompt, tools, or message history changes. Why it matters: If you use tool_choice='any' or tool_choice='tool' in production, requests to Fable 5.1 will fail immediately. If your contract or compliance posture requires zero data retention, you cannot use these models without explicit authorization. The thinking block validation will break conversation replay for new accounts if you modify context mid-conversation. What to do: Grep your codebase for tool_choice parameters set to 'any' or 'tool' and migrate to structured outputs or strict tool use before switching to Fable 5.1. Verify your data retention agreement permits 30-day retention or request authorization from Anthropic if you need zero retention. If you create new accounts after August 31, audit any code that replays thinking blocks after modifying system prompts or tools. https://radar.trulysovereign.ai/items/2caf39c0-b116-452d-8c4f-e5c6119c3081 2caf39c0-b116-452d-8c4f-e5c6119c3081 Mon, 31 Aug 2026 04:10:47 GMT migrate OpenAI deprecations On August 26, 2026, OpenAI announced that whisper-1, gpt-4o-transcribe, gpt-4o-mini-transcribe, and gpt-4o-transcribe-diarize will be removed from the API on February 26, 2027. Recommended replacements are gpt-live-transcribe or gpt-transcribe. Why it matters: Any application using these four models for audio transcription will break on February 26, 2027 unless migrated to the replacement models. You have six months to test and deploy the migration. What to do: Grep your codebase for 'whisper-1', 'gpt-4o-transcribe', 'gpt-4o-mini-transcribe', and 'gpt-4o-transcribe-diarize'. For each usage, schedule migration to gpt-live-transcribe or gpt-transcribe before February 26, 2027. https://radar.trulysovereign.ai/items/c00829cd-4938-43d2-9f3f-f7eaa759fb5b c00829cd-4938-43d2-9f3f-f7eaa759fb5b Sat, 29 Aug 2026 05:59:09 GMT migrate Azure Foundry model retirement schedule MAI-Image-2.5, MAI-Image-2.5-Flash, and MAI-Image-2.5-Pro added to Azure Foundry retirement schedule. All three entered Preview on June 2 or June 19, 2026, and will be retired October 1, 2026. MAI-Image-2e status changed from Preview to Retired. Why it matters: If you are using or planning to use any MAI-Image-2.5 variant, you have a four-month window from preview to retirement. Code written against these models will break on October 1, 2026 unless migrated to a successor model. What to do: Grep your codebase for 'MAI-Image-2.5', 'MAI-Image-2.5-Flash', and 'MAI-Image-2.5-Pro'. If found, schedule migration work to complete before October 1, 2026. If not using these models, no action required. https://radar.trulysovereign.ai/items/cb16fe1f-0d67-4e11-88c8-1290ac610dac cb16fe1f-0d67-4e11-88c8-1290ac610dac Sat, 22 Aug 2026 01:15:30 GMT migrate Claude Platform release notes Released August 20, 2026. Python SDK v1.0 removes the legacy Text Completions API, the temperature/top_p/top_k parameters on Messages methods, and tool runner's client-side compaction_control. Minimum Python version is now 3.10. HTTP layer switches from httpx to httpx2. Async client .with_raw_response results now require await response.parse(). AnthropicBedrock no longer defaults to us-east-1 when AWS region is unconfigured. Why it matters: Code using the removed Text Completions API or deprecated parameters will break immediately on upgrade. Applications running Python 3.9 or earlier cannot install this version. Async response handling that omits await response.parse() will fail. AnthropicBedrock calls without explicit region configuration will error instead of silently using us-east-1. What to do: Before upgrading: grep your codebase for 'text_completions', 'temperature', 'top_p', 'top_k' in Messages calls, and 'compaction_control'; verify Python version is 3.10+; check all .with_raw_response usage includes await response.parse(); confirm AnthropicBedrock has explicit region set. Consult the v1 migration guide for replacement patterns. https://radar.trulysovereign.ai/items/ae29ba19-50c2-440b-8b83-04ca0bf1aaf0 ae29ba19-50c2-440b-8b83-04ca0bf1aaf0 Fri, 21 Aug 2026 01:19:37 GMT migrate Claude Platform release notes The computer use tool is now GA as computer_toolset_20260801, replacing earlier beta versions. The upgrade changes the request shape and tool handling. The Files API is GA with a new response format: file expiration (expires_in_seconds on upload, expires_at in responses) and page/next_page pagination with ids[] filter. Both changes take effect immediately when you stop sending the beta headers. Why it matters: Existing integrations using the beta computer use tool will break if you switch to the GA version without updating your request structure. Files API clients that drop the beta header will receive a different response format and pagination scheme, breaking any code that expects the old structure. What to do: If you use the computer use tool, read the migration guide at 'Migrate from computer_20251124' before removing the beta header. If you use the Files API, test your pagination and expiration handling against the GA format before dropping the files-api-2025-04-14 header from production requests. https://radar.trulysovereign.ai/items/205c4cd5-1cf4-4e6c-8386-ef3c97c4a0a5 205c4cd5-1cf4-4e6c-8386-ef3c97c4a0a5 Thu, 20 Aug 2026 18:39:57 GMT migrate Claude Platform release notes Effective August 19, 2026, three Claude APIs moved from beta to general availability: Files API (/v1/files), Skills API (/v1/skills), and Admin API user-management endpoints. Beta headers (files-api-2025-04-14, skills-2025-10-02, anthropic-beta: ce-user-management-2026-07-13) are no longer required. The Files API GA response format differs from beta: pagination changed from page/next_page to ids[] filter, and file expiration now uses expires_in_seconds (upload) and expires_at (response) instead of the beta format. Why it matters: Code sending beta headers continues to work but receives the old response format. Code that stopped sending headers or never sent them now receives the GA format with different pagination and expiration field names, which will break parsing logic expecting the beta structure. What to do: Grep your codebase for 'files-api-2025-04-14', 'skills-2025-10-02', and 'ce-user-management-2026-07-13'. For Files API calls, verify whether your code parses expires_at and handles ids[] pagination, or if it expects the beta format. If you need the beta format, confirm the header is still being sent; if you need GA format, remove the header and update response parsing. https://radar.trulysovereign.ai/items/ebad5c37-05e6-4d61-96a7-661d9006c8d5 ebad5c37-05e6-4d61-96a7-661d9006c8d5 Thu, 20 Aug 2026 01:15:58 GMT migrate Azure Foundry model retirement schedule Azure Foundry moved gpt-4.1-nano retirement from 2026-10-14 to 2027-04-14. The model's status label changed from 'Deprecated' to 'Legacy' on 2025-04-14. Three other models (gpt-4.1, gpt-4.1-mini, gpt-4o) also changed to 'Legacy' status with 2027-04-14 retirement dates. Why it matters: If you planned migration work around the October 2026 deadline, you now have six additional months. Teams that deprioritized gpt-4.1-nano migration because of the tight timeline can reschedule without urgency. What to do: Check your migration backlog for gpt-4.1-nano work scheduled before 2027-04-14. If you deferred it due to the October deadline, update the target date and reprioritize against other work. https://radar.trulysovereign.ai/items/2a3739cf-3ab9-49a7-ab94-0249afd03331 2a3739cf-3ab9-49a7-ab94-0249afd03331 Wed, 19 Aug 2026 18:36:14 GMT migrate Azure Foundry model retirement schedule Azure added DeepSeek-V4-Flash-0731 to the retirement schedule. The model enters GA on 2026-07-31 and preview access ends on 2026-12-03. Why it matters: If you are testing this model in preview, you have until 2026-12-03 to migrate to the GA version or switch models. Preview endpoints will stop working after that date. What to do: Check if any code or configs reference DeepSeek-V4-Flash-0731 in preview. If so, schedule migration to the GA version before 2026-12-03. https://radar.trulysovereign.ai/items/1a010efc-3b68-4ca7-909c-3aa558516e6b 1a010efc-3b68-4ca7-909c-3aa558516e6b Tue, 18 Aug 2026 01:15:03 GMT migrate Azure Foundry model retirement schedule DeepSeek-R1, MAI-Image-2, Stable-Diffusion-3.5-Large, Stable-Image-Core, and Stable-Image-Ultra moved from Legacy/Preview/Deprecated to Retired status between 31 July and 15 August 2026. MAI-Transcribe-1 added to retirement schedule with cutoff date 15 September 2026. MAI-Transcribe-1.5 added as successor model. Why it matters: Retired models are no longer callable via API. Any code still referencing DeepSeek-R1, MAI-Image-2, or the three Stable models will fail. MAI-Transcribe-1 stops working in 28 days. What to do: Grep your codebase for 'DeepSeek-R1', 'MAI-Image-2', 'Stable-Diffusion-3.5-Large', 'Stable-Image-Core', 'Stable-Image-Ultra', and 'MAI-Transcribe-1'. Remove or replace references before 15 September 2026 for MAI-Transcribe-1; the other five are already dead.