Commit Graph

4969 Commits

Author SHA1 Message Date
Enrico Ros aa441b0656 LLMs: Azure OpenAI: cleaned up (and moved) azureOpenAIAccess - simpler and modularized code 2025-09-12 14:19:20 -07:00
Enrico Ros 39a7e30880 AIX: Azure OpenAI: renamed Env Vars 2025-09-12 13:58:23 -07:00
Enrico Ros 74b69f9ea4 AIX: Azure OpenAI: verified web_search_preview not present yet 2025-09-12 13:57:49 -07:00
Enrico Ros 3094540b93 LLMs: misc bits 2025-09-12 13:24:30 -07:00
Enrico Ros 513500b16e LLMs: extract access response type 2025-09-12 13:23:52 -07:00
Enrico Ros 51c41473a5 Azure: remove path from the example 2025-09-12 12:47:09 -07:00
Enrico Ros e79df4a347 AIX: Azure/Request API: improve web search tool use and exceptions 2025-09-12 11:21:51 -07:00
paulhshort 3d93c856ba Fix Azure OpenAI web_search_preview tool incompatibility
Azure OpenAI doesn't support the web_search_preview tool, which was causing
"Hosted tool 'web_search_preview' is not supported" errors with GPT-5 models.

## Changes:
- Pass dialect information to aixToOpenAIResponses function
- Skip web_search_preview tool addition when dialect is 'azure'
- Add logging when web search is skipped for Azure
- Document known Azure limitations in implementation guide

## Impact:
- Fixes web browsing errors with Azure GPT-5 models
- Maintains web search functionality for regular OpenAI models
- Provides clear logging for debugging

This is a critical fix for Azure OpenAI compatibility as web search is not
currently supported on Azure's Responses API implementation.

🤖 Generated with [Claude Code](https://claude.ai/code)

Co-Authored-By: Claude <noreply@anthropic.com>
2025-08-11 20:07:19 -04:00
paulhshort 9fe5697fd4 Fix Azure OpenAI Resource Not Found errors for GPT-5 and o3 models (#828)
This commit addresses GitHub issue #828 by fixing URL construction for Azure OpenAI's Responses API
and preventing malformed URLs from client configuration issues.

## Problems Fixed:
1. Host normalization: Prevents malformed URLs when client config includes paths/queries
2. API paradigm support: Properly handles Azure's next-gen v1 Responses API
3. API version consistency: Centralizes version management with env overrides

## Key Changes:
- Normalize Azure host URLs to origin only (strip path/query)
- Prefer server environment variables over client-provided hosts
- Add special handling for Responses API (/openai/v1/responses)
- Support both traditional (deployment-based) and v1 API paradigms
- Add configurable API versions via environment variables
- Include debug logging for API paradigm selection

## New Environment Variables:
- AZURE_API_V1: Enable next-gen v1 API explicitly
- AZURE_RESPONSES_API_VERSION: Control Responses API version
- AZURE_CHAT_API_VERSION: Control Chat Completions API version
- AZURE_DEPLOYMENTS_API_VERSION: Control deployments listing API version

## Testing:
Validated with Azure OpenAI endpoint showing:
- List Deployments:  Works
- Chat Completions:  Works (with correct params for GPT-5)
- Responses API (v1):  Works with /openai/v1/responses?api-version=preview
- Responses API (traditional):  404 (Azure doesn't support this pattern)

The fix defaults to using Azure's recommended next-gen v1 API for Responses
while maintaining backward compatibility for existing deployments.

🤖 Generated with [Claude Code](https://claude.ai/code)

Co-Authored-By: Claude <noreply@anthropic.com>
2025-08-11 19:37:41 -04:00
Enrico Ros 7fde3473ea BlobUtils: export direct 2025-08-09 14:12:52 -07:00
Enrico Ros 56a2d68c71 GPT-5: add Markdown Restore option, like the 'o' models 2025-08-08 16:48:11 -07:00
Enrico Ros 3d140604f8 Roll AIX 2025-08-08 16:48:11 -07:00
Enrico Ros 0a2167fa6a AIX: OpenAI: GPT-5 markdown restoration with the Responses API 2025-08-08 16:47:36 -07:00
Enrico Ros 79e95379ec AIX: OpenAI Chat Completions: GPT-5 as an 'o' model 2025-08-08 16:36:40 -07:00
Enrico Ros a4a0ecc0e5 LLMs: reset user params 2025-08-07 19:47:32 -07:00
Enrico Ros 686ad2ed7b FormControls size setting 2025-08-07 19:47:32 -07:00
Enrico Ros d8db79b4e5 AIX: OpenAI does not support Web Search with reasoning:minimal 2025-08-07 14:44:17 -07:00
Enrico Ros d33449f4af GPT-5: support 'minimal' reasoning 2025-08-07 14:37:32 -07:00
Enrico Ros 7e6a12bddf Roll AIX 2025-08-07 14:37:11 -07:00
Enrico Ros ec80413be0 AIX: vnd.oai.reasoning:minimal support 2025-08-07 14:34:54 -07:00
Enrico Ros c2af65facd GPT 5 naming 2025-08-07 13:38:10 -07:00
Enrico Ros 46266ac825 Rename search size to web search 2025-08-07 13:38:10 -07:00
Enrico Ros 91dc25e1c6 OpenAI: GPT5 models (configurable reasoning & search) 2025-08-07 13:38:10 -07:00
Enrico Ros 7f9dafd749 Roll AIX 2025-08-06 15:26:19 -07:00
Enrico Ros 989d0e5741 Anthropic: 4.1 > 4 2025-08-06 15:24:54 -07:00
Enrico Ros 3277c009fa Composer: change tips 2025-08-06 07:57:40 -07:00
Enrico Ros 85f1fe088d Fix models description 2025-08-05 21:37:03 -07:00
Enrico Ros 3c554c92d6 Anthropic: update models 2025-08-05 20:20:42 -07:00
Enrico Ros f95d071197 Anthropic: fix dMessageUtils 2025-08-05 20:20:42 -07:00
Enrico Ros da887d58db Incognito: improve chat drawer icon 2025-08-05 20:20:42 -07:00
Enrico Ros 5273a3c84f ClosablePopup: support size 2025-08-05 20:13:38 -07:00
Enrico Ros f51712867f Folders: feed items count (commented) 2025-08-05 20:13:38 -07:00
Enrico Ros ecac1dffec Folders: support for items count, if any 2025-08-05 20:13:37 -07:00
Enrico Ros 28817bee72 Folders: only drag on the icon 2025-08-05 20:04:04 -07:00
Enrico Ros 3fd41329ea ChatDrawer: improve stepping of the progressive limiter 2025-08-05 20:04:04 -07:00
Enrico Ros f734f0b5f7 ChatDrawer: limit items (ListItems are on the heavy side) 2025-08-04 23:55:37 -07:00
Enrico Ros 3b34a8b96d Fix blob conversion types with newer TS 2025-08-04 14:19:52 -07:00
Enrico Ros 74e6ee4b2d Posthog: improve config (if/when enabled) 2025-08-04 14:15:45 -07:00
Enrico Ros a3e216c956 Logger: max 200 entries 2025-08-02 19:20:37 -07:00
Enrico Ros d4203f728e Nicer replacement of ph when interrupted 2025-08-01 15:34:11 -07:00
Enrico Ros 616376f4ac BackupRestore: Flash: vendor prefix 2025-08-01 14:45:49 -07:00
Enrico Ros 1a309c9bdf BackupRestore: Flash: change schema/version 2025-08-01 14:36:07 -07:00
Enrico Ros 253fc3b213 Approximate Tokenization - optimized 2025-08-01 14:27:05 -07:00
Enrico Ros a79fd0a10c Approximate Tokenization - default on new Mobile installs 2025-08-01 14:26:26 -07:00
Enrico Ros 04df3dcba8 Butter-smooth decimator with deadlines 2025-07-31 11:33:13 -07:00
Enrico Ros 00fbf77dbd Raise update frequency due to optimizations 2025-07-31 09:07:47 -07:00
Enrico Ros 9a34c1e376 References (to Image Assets): restore places that assumed 'image_ref' 2025-07-30 18:06:40 -07:00
Enrico Ros e248104d4b Reference: embed a ref summary 2025-07-30 17:11:22 -07:00
Enrico Ros c10558f230 Reference: cleanup migrated fragments 2025-07-30 16:26:41 -07:00
Enrico Ros 5be41b8199 References: add support (migration, creation (attaching, aix.reassembler, t2i , ego-msg), gc dblobs collection, render image, render button icon, dereference, bucketing, token count) with exhaustive checks 2025-07-30 16:09:05 -07:00