maximhq/bifrost ent-v2.0.0-prerelease3-base

Enterprise v2.0.0-prerelease3 base

on GitHub

What's Changed

• helm update by @akshaydeo in #5216
• fix changelog for audit log s3 storage by @akshaydeo in #5219
• fix changelog for audit log s3 storage (#5219) by @akshaydeo in #5220
• docs(alerting): add alerting documentation by @SahilChoudhary22 in #4442
• docs(alerting): add UI screenshots and fix CEL variable accuracy by @SahilChoudhary22 in #5232
• feat: add dual_credential_conflict_behavior to resolve IDP token + VK conflicts on inference requests by @BearTS in #5201
• feat: add generic OIDC SCIM/SSO provider support to Helm chart by @BearTS in #5173
• feat: add dual_credential_conflict_behavior config option with UI selector for dual-auth conflict handling by @BearTS in #5235
• feat: add bifrost.client.dualCredentialConflictBehavior to Helm chart by @BearTS in #5227
• feat: add support for request headers to be set using helm in otel profiles by @BearTS in #5236
• feat: add provisioningToken and claimScimAttributes to SCIM providers for declarative GitOps provisioning in helm by @BearTS in #5240
• chore: release Bifrost Helm chart v2.1.29 by @BearTS in #5241
• docs update by @akshaydeo in #5299
• updates claude code anthropic version enforcement by @akshaydeo in #5347
• fix: complete visible thinking items on the streaming Responses surface by @fus3r in #5150
• fix: resolve merge conflict in TableClientConfig by @impoiler in #5349
• fix: adds sgl anthropic compatible apis via key level setting by @sammaji in #5305
• fix: uses anthropic endpoints in chat completion and responses in deepseek by @sammaji in #5176
• chore: update deepseek provider docs by @sammaji in #5177
• chore: adds deepseek to provider harness api tests by @sammaji in #5208
• fix: enable raw request and store for mistral ocr requests by @sammaji in #5310
• fix: adds count tokens handling for deepseek and sgl by @sammaji in #5311
• fix: moves vllm to native responses api by @sammaji in #5306
• fix: adds support for anthropic apis in fireworks by @sammaji in #5332
• fix: drops unsupported reasoning summary values for azure model router by @sammaji in #5242
• fix: fallback select shows trimmed model names by @sammaji in #5257
• fixes query creation from user input by @akshaydeo in #5357
• [Fix]: OpenAI provider - default detial to auto on input_image blocks by @AdityaPainuli in #5352
• refactor: make RefineModelForProvider idempotent and unify Groq/Replicate/Perplexity into a single case by @Pratham-Mishra04 in #5133
• feat: wafer ai provider by @TejasGhatte in #5222
• docs: clarify that Level 1 LB always runs for pinned providers, appending healthy fallbacks instead of skipping by @Pratham-Mishra04 in #5135
• docs: add append_fallbacks_to_pinned as a fifth load-balancer switch and update all references to pinned-request fallback behavior by @Pratham-Mishra04 in #5336
• docs: add Response Lifecycle endpoints and expand streaming event types table for OpenAI Responses API by @Pratham-Mishra04 in #5343
• docs: add retrieve, delete, cancel, and list input items endpoints for stored responses by @Pratham-Mishra04 in #5344
• docs: add response.web_search_call.in_progress and response.web_search_call.completed to stream response type enums by @Pratham-Mishra04 in #5345
• docs: expand output field schema in OpenAPI from generic object to typed oneOf (string, content blocks, computer screenshot) by @Pratham-Mishra04 in #5350
• feat: add toggle for always retain content in object storage by @roroghost17 in #5370
• feat: add appendFallbacksToPinned load-balancer option to Helm chart by @Pratham-Mishra04 in #5374
• feat: validate routing rule scope_id resolves to an existing entity at write time and during config.json sync by @Pratham-Mishra04 in #5371
• feat: add bifrost.webhooks and bifrost.client.webhookConfig Helm values with schema validation by @Pratham-Mishra04 in #5375
• feat: add keep_alive_timeout_in_seconds to bifrost network config schema and helm chart by @Pratham-Mishra04 in #5376
• docs: wafer ai provider by @TejasGhatte in #5369
• chore: update helm charts with retain_content_in_object_storage by @roroghost17 in #5372
• fix: exclude unresolved VKs from scope ID set and normalize empty routing rule scope to global by @Pratham-Mishra04 in #5377
• fix: convert structured output format to tool for Azure provider in Anthropic chat and responses requests by @Pratham-Mishra04 in #5379
• fix: use time.Now().Add instead of channel receive for deadline and bypass webhook check in nil-dispatcher test by @Pratham-Mishra04 in #5381
• test: extend structured-output tool-conversion tests to cover Vertex, BedrockMantle, and Azure providers by @Pratham-Mishra04 in #5382
• test: add Azure Claude structured output (json_schema) e2e test by @Pratham-Mishra04 in #5383
• fix: complete Cohere streaming Responses terminal events with the output array and reasoning payloads by @fus3r in #5386
• docs:add docs for creating management api keys by @R-droid101 in #5342
• fix: anthropic fallbacks and refusal by @TejasGhatte in #5367
• fix: bill for serving fallback model in anthropic by @TejasGhatte in #5387
• docs: add content-logging docs by @roroghost17 in #5389
• fix: bedrock reasoning double emission by @TejasGhatte in #5388
• fix: use the central method for content attribute check in OTEL by @roroghost17 in #5393
• feat: show server fallback model in logs by @TejasGhatte in #5391
• chore: fixes migration tests by @sammaji in #5390
• feat: inline error/error_omitted fields in webhook payloads for failed async jobs by @Pratham-Mishra04 in #5398
• docs: add data.error and data.error_omitted webhook fields and update max_response_payload_kbs description by @Pratham-Mishra04 in #5399
• chore: remove WEBHOOK_EVENT_COLORS and sort events in webhook badges by @Pratham-Mishra04 in #5400
• feat: add content disabled message on logs UI by @roroghost17 in #5401
• fix: gate matview read path on shape check to prevent "column does not exist" during rolling deploys by @impoiler in #5384
• feat: track Anthropic extended-thinking tokens as ReasoningTokens across chat, responses, and passthrough by @impoiler in #5397
• chore: release Bifrost Helm chart v2.1.30 by @BearTS in #5402
• chore: disable Docker Hub Helm chart push step by @BearTS in #5403
• feat(helm+docs): add stream_replay_duration_ms to guardrail rules for block-capable streaming output replay control by @Madhuvod in #5253
• core gate changes for streaming replay client delivery by @Madhuvod in #5392
• changelogs by @akshaydeo in #5404
• adds 503 for request dropped by @akshaydeo in #5405
• drop attachments from last message when hybrid mode is on by @akshaydeo in #5406
• cherrypick anyOf, oneOf change back in dev by @akshaydeo in #5408
• test case fixes by @akshaydeo in #5411
• adds domain whitelists by @akshaydeo in #5413
• v1.6.5 by @akshaydeo in #5414
• fix(logstore): rebuild ContentSummary from S3 after hydration to prevent encoding corruption by @lvsao in #5415
• fixed version bumps by @akshaydeo in #5417
• redundant line cleanup in helm readme by @Madhuvod in #5418
• fixed version bumps by @akshaydeo in #5419
• chore: fix migration tests by @sammaji in #5420
• framework test fixes by @akshaydeo in #5421
• chore: adds docs for azure model router by @sammaji in #5175
• fix: adds created timestamp for openai text completions by @sammaji in #5407
• test fixes --skip-ci by @akshaydeo in #5422
• docs update by @akshaydeo in #5423
• docs update by @akshaydeo in #5424
• fixes helm changelog (#5426) by @akshaydeo in #5427
• docs: clarify x-bf-async-webhook opt-in delivery by @Pratham-Mishra04 in #5430
• docs: update webhooks next steps links for virtual keys and logs_store by @Pratham-Mishra04 in #5431
• release notice by @akshaydeo in #5445
• fixed notice by @akshaydeo in #5446
• fixed error descirption for inference jwt token by @akshaydeo in #5447
• updated message for JWT issue by @akshaydeo in #5449
• docs: remove AWS ALB from identity-aware proxy changelog entries by @BearTS in #5453
• updated error messages by @akshaydeo in #5456
• docs: add auth proxy docs for Cloudflare ZTNA and generic OIDC proxy by @BearTS in #5460
• docs: remove mermaid diagrams and simplify warnings in auth proxy docs by @BearTS in #5461
• feat: redesign models catalog with mode filter, cost columns, pagination, and improved model detail dialog by @impoiler in #5479
• fix: models-catalog/list docs page by @impoiler in #5480
• fix: correct rendering on the models-catalog/list docs page by @impoiler in #5481
• opus-5 cut by @akshaydeo in #5548
• enterprise changelog by @akshaydeo in #5549
• fix: return closed non-nil channel for empty streams instead of (nil, nil) by @kharkevich in #5556
• override count fixes for multinode setup by @akshaydeo in #5558
• fixes live reload nuking model list before sync by @akshaydeo in #5559
• test fixes by @akshaydeo in #5569
• adds count token api support for bedorck mantle by @akshaydeo in #5563
• adds ssetruncation interface for handling EOF cases by @akshaydeo in #5566
• adds EOF handler support to all providers by @akshaydeo in #5567
• fix done marker for error frame by @akshaydeo in #5568
• feat: add customer_id/customer_ids attachment and UpdateTeamResponse schema to updateTeam by @impoiler in #5574
• feat: drive model limit edit sheet from URL query param with fallback fetch for off-page configs by @impoiler in #5577
• docs: responses retrieve stream by @TejasGhatte in #5570
• fix: IncludeServerSideToolInvocations field in genai api by @TejasGhatte in #5576
• fix: follow redirects for download in genai passthrough by @TejasGhatte in #5579
• fix: convert strict null to false for sdk compatibility by @TejasGhatte in #5587
• feat: add partitioned sidekiq job claiming with FIFO ordering per key by @BearTS in #5572
• fix: vllm stream truncation by @sammaji in #5585
• fix: copy url for skills repo marketplace fix by @danpiths in #5591
• models list backend changes by @akshaydeo in #5561
• model list refresh UI changes by @akshaydeo in #5562
• chore: update flake by @roroghost17 in #5537
• chore: refactor overhead duration export for connectors by @roroghost17 in #5594
• fix: apply ScopedDB to FindMCPToolLog for query-scope enforcement by @Pratham-Mishra04 in #5580
• feat: add user-scoped pricing override family (user, user_provider, user_provider_key) by @Pratham-Mishra04 in #5581
• feat: add user-scoped pricing overrides with pluggable enterprise user picker by @Pratham-Mishra04 in #5582
• feat: add user-scoped routing rules with user_id CEL variable support by @Pratham-Mishra04 in #5583
• feat: add user scope to routing rules with pluggable user picker registry by @Pratham-Mishra04 in #5584
• fix: bedrock mantle set structured outputs to false for claude models by @TejasGhatte in #5592
• fix: merge additional tools into tools for bedrock mantle by @TejasGhatte in #5593
• refactor: update the budget & rate limits page for text changes by @roroghost17 in #5589
• refactor: update DD docs location by @roroghost17 in #5590
• chore: migration test fixes by @TejasGhatte in #5607
• refactor: extract shared VirtualKeySelector component and replace inline VK query logic across pricing, routing, MCP, and model-limit surfaces by @impoiler in #5614
• feat: replace static list queries with async entity selectors for teams, customers, and virtual keys by @impoiler in #5616
• feat: replace full VirtualKeys preload with virtual_key_count on paginated customer list by @impoiler in #5618
• fix: show FullPageLoader on initial fetch to prevent empty-state flash in pricing overrides and model limits views by @impoiler in #5623
• feat: add RankingLimit filter with all/limit query params to cap or remove ranking row limits by @impoiler in #5624
• feat: add limit/all params to rankings endpoints and use uncapped export snapshots in PDF/CSV by @impoiler in #5625
• docs: add entity selector guidelines to AGENTS.md by @impoiler in #5635
• feat: add shared ProviderConfigCard component and budgetOutline helpers by @BearTS in #5636
• refactor: extract provider config accordion into reusable ProviderConfigCard component by @BearTS in #5637
• feat: add per-tab export scope to dashboard CSV and PDF export by @impoiler in #5642
• refactor: replace preloaded customers/virtual-keys with CustomerSelector and server-computed virtual_key_count in teams view by @impoiler in #5643
• feat: add CopyableId component to customer, team, and virtual key detail sheets by @impoiler in #5586
• fix: add customer_id/business_unit_id to matview scopeProjection so team-data DAC scope resolves without column errors by @impoiler in #5608
• fix: scope filterdata cache per caller and add item-budget admission control by @impoiler in #5609
• accepts output_text user blocks in complexity extraction by @kohlivrinda in #5600
• lifts chat-completions function wrapper in responses tool unmarshal by @kohlivrinda in #5601
• fixes skipped cases in semanticcache tests by @sammaji in #5605
• refactor: replace preloaded teams/customers with TeamSelector/CustomerSelector in virtual keys table and sheet by @impoiler in #5644
• fixes skipped framework tests by @sammaji in #5611
• replaces deprecated huggingface model from core tests by @sammaji in #5615
• otel adds export timeouts by @akshaydeo in #5610
• control db connections for all cases by @akshaydeo in #5612
• batch writing for dump budgets and dump ratelimits by @akshaydeo in #5632
• feat: expose streaming request types as distinct _stream suffixed values in routing CEL by @Pratham-Mishra04 in #5603
• fix: vertex count tokens unsupported fields by @TejasGhatte in #5619
• fix: count tokens request building by @TejasGhatte in #5620
• fix: image search for genai search tool by @TejasGhatte in #5647
• feat: add user_id filter to virtual keys list (enterprise-only, OSS fails closed) by @impoiler in #5653
• chore: harness tests count tokens by @TejasGhatte in #5648
• fix: add no-padding-parent and consistent padding across workspace pages by @impoiler in #5660
• fix: set octet-stream Content-Type on pprof endpoints for scraper compatibility by @tcx4c70 in #5071
• chore: release Bifrost Helm chart v2.1.32 by @BearTS in #5540
• feat(vertex): allow Flex PayGo for gemini-3.5-flash-lite by @gstvortiz1 in #5598
• docs: clarify SDK virtual key placeholders by @R-droid101 in #5664
• fix: include CachedWriteTokenDetails in prompt token details response by @impoiler in #5666
• feat: add SyncRoutingRules for atomic batch routing rule sync with deferred priority check by @BearTS in #5678
• recalculate cost fixes by @akshaydeo in #5669
• required text for document calls fix by @akshaydeo in #5675
• test case fixes by @akshaydeo in #5681
• changelog updates by @akshaydeo in #5665
• adds CalculateCost and GetModelConfig for ctx in plugins by @TejasGhatte in #5682
• v1.6.7 by @akshaydeo in #5685
• schema fixes by @akshaydeo in #5688
• schema test fixes by @akshaydeo in #5689
• docs by @akshaydeo in #5692
• fixes version suffix for go plugins docs by @akshaydeo in #5697
• plugin docs fixes by @akshaydeo in #5698
• chore: release Bifrost Helm chart v2.1.33 with OTEL schema fix by @BearTS in #5705
• docs: add budget override API reference and OpenAPI spec by @impoiler in #5736
• docs: add per-user access-profile budget override API and docs by @impoiler in #5747
• add proxy support for realtime calls (websocket based) by @akshaydeo in #5788
• fix: inject placeholder text block for document-only messages in Bedrock Converse API by @Pratham-Mishra04 in #5817
• retain tool cache_control for bedrock converse by @akshaydeo in #5811
• bedrock invoke flow now retains image/tool_use/tool_return by @akshaydeo in #5814
• fixes #5186 - encrypted reasoning by @akshaydeo in #5819
• bedrock + anthropic patches for adaptive thinking by @akshaydeo in #5821
• cache_control for bedrock by @akshaydeo in #5824
• stream cancel fix by @akshaydeo in #5836
• closed channel fix by @akshaydeo in #5837
• fix: tool ordering in extractToolsFromResponsesConversationHistory by @TejasGhatte in #5828
• fix: bedrock headers signing denylist by @TejasGhatte in #5833
• perf: bulk VK provider replacement and direct VK lookup by @danpiths in #5844
• updates codeowners by @akshaydeo in #5845
• feat: add WithFasthttpBufferSizes option to HTTPClientFactory for configurable SCIM read/write buffers by @BearTS in #5808
• fix: tolerate ErrNotFound when pruning cascade-deleted budgets and configs by @BearTS in #5802
• provider harness fixes by @akshaydeo in #5846
• chore: adds claude mcpb docs by @sammaji in #5818
• chore: adds docs for github copilot cli, app and vscode extension by @sammaji in #5830
• moves sse hearbeats to a common structure to reuse by @akshaydeo in #5850
• fix: adds context_management param in responses api by @sammaji in #5750
• feat: surface external (access-profile) rate limit in VK quota and admin read paths by @BearTS in #5857
• feat: add is_access_profile_managed server-computed flag to VK for RBAC-independent managed-key UI by @BearTS in #5858
• token parity tests by @akshaydeo in #5856
• docs: add AWS ALB identity-aware proxy guide and update role resolution strategy docs by @BearTS in #5826
• template fixes by @akshaydeo in #5860
• chore: release Bifrost Helm chart v2.1.34 with SCIM roleResolutionStrategy support by @BearTS in #5777
• docs by @akshaydeo in #5866
• v1.6.9 changes by @akshaydeo in #5963
• docs and test case improvements by @akshaydeo in #5965
• encrypted content bug fix by @akshaydeo in #5982
• fixes cohere issue by @akshaydeo in #5983
• gemini reasoning fixes by @akshaydeo in #5984
• improved sdk integration testss by @akshaydeo in #6008
• core release by @akshaydeo in #6010
• fixed gates for core builds by @akshaydeo in #6030
• enterprise core version update by @akshaydeo in #6031
• quarterly budget by @akshaydeo in #5996
• governance: quarterly budget windows with configurable fiscal start by @akshaydeo in #5997
• ui: quarterly budget period with fiscal-quarter advanced settings by @akshaydeo in #5999
• docs: document quarterly budgets and fiscal quarter configuration by @akshaydeo in #6000
• adds reset budget honor flow by @akshaydeo in #6002
• governance: calendar alignment applies from the next period by @akshaydeo in #6003
• governance: extend reset_budget_usage to teams, customers, model limits and provider governance by @akshaydeo in #6004
• governance: enabling calendar alignment preserves the open window by @akshaydeo in #6013
• fix: normalize Together provider for pricing lookup by @dani29 in #6037
• feat: add per-model budgets/rate-limits to virtual key provider configs by @BearTS in #5703
• fix: show model budget count in provider card header summary by @BearTS in #5704
• feat: replace per-budget BudgetOverrideDialog with unified BudgetOverrideManagerDialog grouping provider and model budgets by @BearTS in #5898
• v1.5.10 by @akshaydeo in #6076
• feat(logstore): fan out team/customer/BU dimension reads to JSON-array columns on all dialects by @impoiler in #6092
• docs: upgrade postgres image to 16-alpine in on-premise guide by @impoiler in #6094
• harness fixes by @akshaydeo in #6088
• fixed harness issue by @akshaydeo in #6107
• changelogs by @akshaydeo in #6108
• core updates by @akshaydeo in #6110
• docs update by @akshaydeo in #6122
• fix: map Bedrock content_filter/guardrail_intervened stop reasons to incomplete status with content_filter incomplete details by @BearTS in #6103
• adds path for skipping auth by @akshaydeo in #6124
• feat(runware): async 3D generation via /videos + passthrough route by @R-droid101 in #6075
• updates skip-core-test flag and changelog by @akshaydeo in #6128
• v2.0.0-prerelease3 by @akshaydeo in #6127
• chore: release Bifrost Helm chart v2.1.35 with MCP OAuth, setup token, and governance enhancements by @BearTS in #6129
• fix: updates framework tests by @sammaji in #6134
• fix: force TCP for postgres pg_isready checks by @sammaji in #6135
• fix: correct missing fi closings and remove duplicate redaction_mapping block in migration test script by @BearTS in #6133
• feat: expose resource (RFC 8707) in oauth_config schema and mark flattened oauth_* response fields as excluded from config sync by @Pratham-Mishra04 in #6136
• fix(governance): block from_memory VK reads for DAC-scoped callers … by @akshaydeo in #6138
• fix: exclude PKCE/state/token_id columns from oauth_configs snapshot comparison and fix isPayloadEmpty to check only offloadable payload fields by @BearTS in #6139
• fix(gemini): prefer function declarations over google search when tool combination is disabled by @impoiler in #6065
• fix(gemini): allow Vertex AI to send mixed tools without the server-side invocations flag by @impoiler in #6066
• feat(gemini): support server-side toolCall/toolResponse parts with thoughtSignature round-trip fidelity by @impoiler in #6071

New Contributors

• @lvsao made their first contribution in #5415
• @kharkevich made their first contribution in #5556
• @kohlivrinda made their first contribution in #5600
• @gstvortiz1 made their first contribution in #5598

Full Changelog: ent-v2.0.0-prerelease2-base...ent-v2.0.0-prerelease3-base