v1.15.2

What's Changed

• Qualifire | Update Guardrails by @yuval-qf in #1483
• feat: update readme by removing 2024 report and AI engineering hours by @siddharthsambharia-portkey in #1493
• fix: support azure blob for batches by @b4s36t4 in #1496
• fix: allow custom scope work with azure agents by @b4s36t4 in #1495
• make anthropic beta header usage consistent by @narengogi in #1491
• [bug/behaviour] only execute enabled checks when sequential guardrails are enabled by @narengogi in #1492
• fix: Bump Vercel Cookbook package versions by @beast-nev in #1497
• 1.15.2 by @VisargD in #1499

New Contributors

• @beast-nev made their first contribution in #1497

Full Changelog: v1.15.1...v1.15.2

v1.15.1

What's Changed

• fix: gemini minimal reasoning effort by @narengogi in #1486
• feat: azure checks for protected material and sheild prompt by @b4s36t4 in #1480
• fix: use correct url for batches by @b4s36t4 in #1488
• 1.15.1 by @VisargD in #1489

Full Changelog: v1.15.0...v1.15.1

v1.15.0

What's Changed

• Thinking level support for Gemini 3 Pro by @maciej-nexos in #1444
• anthropic models on azure by @narengogi in #1456
• Add IO Intelligence provider support by @eminaruk in #1451
• feat: gemini 3 in openrouter by @stefan-jiroveanu in #1454
• fix(docs): broken links by @juzhiyuan in #1470
• feat(provider): add AI Badgr OpenAI-compatible provider by @miguelmanlyx in #1472
• concatenate multple text parts for gemini models by @narengogi in #1466
• feat: Add support for Anthropic advanced tool use beta features by @roh26it in #1452
• feat: support azure v1 api & send deployment-name for responses by @b4s36t4 in #1467
• allow anthropic beta header for all routes on anthropic provider by @narengogi in #1463
• oracle provider by @narengogi in #1461
• add hallucination eval by @joshweimer-patronusai in #1434
• feat: add openai compatible response support for anthropic on azure by @arturfromtabnine in #1465
• improvements for nano banana by @narengogi in #1476
• fix(anthropic): add support for tool_choice=none by @ShivamB25 in #1477
• feat: add thinking parameter to OllamaChatCompleteConfig by @siddharthsambharia-portkey in #1415
• fix: type issues with transform type update by @b4s36t4 in #1479
• feat: add Not Null plugin to check for null or empty response content by @Samveg12 in #1481
• feat: add support for google maps grounding by @b4s36t4 in #1478
• Add OVHcloud AI Endpoints provider by @eliasto in #1471
• execute guardrails sequentially by @narengogi in #1475
• 1.15.0 by @VisargD in #1482

New Contributors

• @maciej-nexos made their first contribution in #1444
• @eminaruk made their first contribution in #1451
• @stefan-jiroveanu made their first contribution in #1454
• @juzhiyuan made their first contribution in #1470
• @miguelmanlyx made their first contribution in #1472
• @joshweimer-patronusai made their first contribution in #1434
• @ShivamB25 made their first contribution in #1477
• @Samveg12 made their first contribution in #1481
• @eliasto made their first contribution in #1471

Full Changelog: v1.14.3...v1.15.0

v1.14.3

What's Changed

• fix google throught signature by @narengogi in #1450
• 1.14.3 by @VisargD in #1457

Full Changelog: v1.14.2...v1.14.3

v1.14.2

What's Changed

• fix headers in streaming response for cached results by @narengogi in #1447
• 1.14.2 by @VisargD in #1448

Full Changelog: v1.14.1...v1.14.2

v1.14.1

What's Changed

• feat: add Z_AI provider by @siddharthsambharia-portkey in #1416
• feat: add modal labs provider and related header keys to mask while logging by @roh26it in #1419
• send empty usage object in message start chunk for bedrock messages by @narengogi in #1424
• handle falsy check in vertex parameter mapping by @narengogi in #1422
• feat: Add Support for Additional Audio Formats in Vertex AI Provider by @roh26it in #1418
• send hook results when streaming response by @narengogi in #1407
• fix: return null for empty tools in transformToolsConfig by @siddharthsambharia-portkey in #1429
• simplify openai base transform for chat complete by @narengogi in #1420
• simplify pr template by @narengogi in #1431
• do not forwards headers for all providers by @narengogi in #1435
• Auto add anthropic headers by @ayush-portkey in #1433
• remove region identifier prefix when making request to count tokens endpoint by @narengogi in #1436
• feat: add f5-guardrail plugin by @b4s36t4 in #1440
• thought messages support for gemini 3 pro by @narengogi in #1443
• 1.14.1 by @VisargD in #1446

Full Changelog: v1.14.0...v1.14.1

v1.14.0

What's Changed

• handle malformed guardrails and add stack trace to chat completions handler by @narengogi in #1388
• fix: add integrationDetails and virtualKeyDetails to constructConfigF… by @VisargD in #1391
• #1300 | Add Qualifire integration by @yuval-qf in #1311
• improvement: added reasoning_effort for grok by @marsianin in #1396
• Fix : Walled Ai Guardrail changed to walledprotect by @indranil786 in #1316
• add newer parameters to respones by @narengogi in #1403
• [Snyk] Fix for 4 vulnerabilities by @roh26it in #1380
• fix: add role 'assistant' to chat completion stream chunk by @siddharthsambharia-portkey in #1400
• fix: handle error from bedrock batch status check by @b4s36t4 in #1369
• Rate limits and budgets by @narengogi in #1366
• migrate to v2 for cohere embed and chat by @narengogi in #1384
• computer use for google by @narengogi in #1401
• feat: add matterai provider by @code-crusher in #1393
• feat: add support for linked Security Profile API keys, additional tests by @jroberts2600 in #1398
• Add performance monitoring (prometheus for metrics and winston for logging) by @narengogi in #1394
• [vertex-ai] fix: allow object as tool content by @arturfromtabnine in #1279
• Feat/azure workload identity by @b4s36t4 in #1408
• improvement: add cache details to usage for (vertex ai) gemini and anthropic models by @miuosz in #1373
• feat: support claude code oauth/max plans on Portkey by @siddharthsambharia-portkey in #1381
• fix: custom host validator to protect from ssrf attack by @sk-portkey in #1372
• fix: return Response instead of object, typescript by @b4s36t4 in #1409
• fix: add types and checks for builds by @b4s36t4 in #1411
• fix: build errors by @VisargD in #1410
• revert apm changes to avoid warninglogs by @narengogi in #1412
• 1.14.0 by @VisargD in #1413

New Contributors

• @yuval-qf made their first contribution in #1311
• @marsianin made their first contribution in #1396

Full Changelog: v1.13.0...v1.14.0

v1.13.0

What's Changed

• feat: add javelin guardrails by @abhijitjavelin in #1332
• handle multiple checks of the same type in a hook by @narengogi in #1377
• feat: add CometAPI provider integration by @tensornull in #1364
• chore: handle object response from pre request validator function by @VisargD in #1382
• use invoke model to count tokens for anthropic through bedrock by @narengogi in #1379
• add input_audio parameter support for vertex by @narengogi in #1367
• improvement: forward cache control parameter for plain text content (system messages) by @miuosz in #1374
• bedrock provider cleanup by @narengogi in #1276
• fix falsy check in bedrock by @narengogi in #1383
• add some parameters mapping by @narengogi in #1349
• cleanup vertex api.ts by @narengogi in #1350
• fix type issue by @narengogi in #1385
• cleanup azure openai implementation by @narengogi in #1352
• support api key for bedrock by @narengogi in #1295
• handle streaming for transcription endpoint by @narengogi in #1376
• New "Add Prefix" Guardrail by @vrushankportkey in #1258
• fix destructuring in vertex-ai/index by @narengogi in #1386
• 1.13.0 by @VisargD in #1387

New Contributors

• @abhijitjavelin made their first contribution in #1332
• @tensornull made their first contribution in #1364
• @miuosz made their first contribution in #1374

Full Changelog: v1.12.3...v1.13.0

v1.12.3

What's Changed

• anthropic beta for vertex-ai by @narengogi in #1356
• fix: add tool_use, thinking and redacted_thinking content_block_start event handling in Bedrock messages transformer by @VisargD in #1362
• 1.12.3 by @VisargD in #1365

Full Changelog: v1.12.2...v1.12.3

v1.12.2

What's Changed

• fix: tools cache_control parameter mapping for bedrock unified messages request transformer by @VisargD in #1360
• 1.12.2 by @VisargD in #1361

Full Changelog: v1.12.1...v1.12.2