diff --git a/.release-please-manifest.json b/.release-please-manifest.json
index 76d5538a4..7d9b009db 100644
--- a/.release-please-manifest.json
+++ b/.release-please-manifest.json
@@ -1,3 +1,3 @@
 {
-  ".": "0.9.0"
+  ".": "0.10.0"
 }
diff --git a/CHANGELOG.md b/CHANGELOG.md
index 1603825ed..7a63425d3 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -1,5 +1,200 @@
 # Changelog
 
+## [0.10.0](https://github.com/superlinked/sie/compare/v0.9.0...v0.10.0) (2026-10-09)
+
+
+### ⚠ BREAKING CHANGES
+
+* **server:** a LoRA whose PEFT config sets bias to "all" or "lora_only" no longer loads. A profile that declares one still loads its model without that LoRA and logs the refusal. A request that names one gets the same retryable LORA_LOADING response as for any other LoRA that fails to load in the background, and the server log carries the refusal. Retrain or export the LoRA with bias="none".
+
+### Features
+
+* **config:** accept hybrid encode and score routing in a cluster ([#576](https://github.com/superlinked/sie/issues/576)) ([96de71c](https://github.com/superlinked/sie/commit/96de71cd2d2ef977c0003eae8c961567efd904a5)), closes [#415](https://github.com/superlinked/sie/issues/415)
+* **config:** refuse a remote profile that names an undefined upstream ([#500](https://github.com/superlinked/sie/issues/500)) ([4977097](https://github.com/superlinked/sie/commit/497709744ce5e827aa89a8191a840c6ea14b65ce))
+* **examples:** add bounded agent stage latency trials ([2326f19](https://github.com/superlinked/sie/commit/2326f19b2b9a04670b7d355e72085c4b651c0842))
+* **examples:** add custom-entity-types, caller-named entity types against LLMs, Comprehend and spaCy ([#468](https://github.com/superlinked/sie/issues/468)) ([7860d1f](https://github.com/superlinked/sie/commit/7860d1f28e17c899db8f011ae8b4d5ccb76ab8c1))
+* **examples:** add reproducible agent stage latency evaluation ([aa20315](https://github.com/superlinked/sie/commit/aa20315ba13ac92221d3fdb348b1183f644114d4))
+* **examples:** add reproducible photo transcription study ([#630](https://github.com/superlinked/sie/issues/630)) ([a826a95](https://github.com/superlinked/sie/commit/a826a9554de22dfcd420a7fa20c2dd073342b342))
+* **examples:** add structured-output-accuracy, strict structured outputs scored on the Structured Output Benchmark ([#474](https://github.com/superlinked/sie/issues/474)) ([96b7f9b](https://github.com/superlinked/sie/commit/96b7f9b63d9035853851cc901fc2e146e2ad62ba))
+* **examples:** compare matched short-verdict guardrails configurations ([#479](https://github.com/superlinked/sie/issues/479)) ([3ebda81](https://github.com/superlinked/sie/commit/3ebda81134cbba29330f826f7b946b8dfca32fe1))
+* **examples:** document-to-markdown-olmocr prints the score without the math files ([#459](https://github.com/superlinked/sie/issues/459)) ([37e1380](https://github.com/superlinked/sie/commit/37e13808c412f851f7517247dad0907b0d4185d8))
+* **examples:** evaluate ordinary named entity tags ([#629](https://github.com/superlinked/sie/issues/629)) ([4ecec8a](https://github.com/superlinked/sie/commit/4ecec8ad13c6ebb620f88ad74c633ac061a02d88))
+* **examples:** rebuild guardrails as a harmful-prompt screening study ([#456](https://github.com/superlinked/sie/issues/456)) ([01e606e](https://github.com/superlinked/sie/commit/01e606e616e09bf59bbcb7c0d6ecc165f222399f))
+* **examples:** rebuild image-search on a held-out Amazon Berkeley Objects product catalogue ([#470](https://github.com/superlinked/sie/issues/470)) ([8041b15](https://github.com/superlinked/sie/commit/8041b15b47ec44d92b671fc8f6c6b711bad79312))
+* **examples:** rebuild redact on the pre-registered PII coverage run ([#452](https://github.com/superlinked/sie/issues/452)) ([f99850a](https://github.com/superlinked/sie/commit/f99850a7b66905b66977c15af5d189a60e1aa783))
+* **examples:** rebuild visual-document-search on the ViDoRe v3 head-to-head ([#472](https://github.com/superlinked/sie/issues/472)) ([832f561](https://github.com/superlinked/sie/commit/832f5612a760c208275bc3541f3e3573e171432c))
+* **examples:** replay document-field extraction evidence ([#511](https://github.com/superlinked/sie/issues/511)) ([971f2d4](https://github.com/superlinked/sie/commit/971f2d4f55684ae1d963c19bbb43e8c3399d35aa))
+* **examples:** reproduce caller-rule rerank study ([#485](https://github.com/superlinked/sie/issues/485)) ([c5df8f9](https://github.com/superlinked/sie/commit/c5df8f98d2ec3f840a1bf7ed9658d2417ab8365a))
+* **examples:** reproduce GLM-OCR receipt and handwriting results ([#486](https://github.com/superlinked/sie/issues/486)) ([ea15097](https://github.com/superlinked/sie/commit/ea150976ef272ba5d3ab9b70778643c5451c9e4e))
+* **examples:** reproduce RF100-VL named-object detection ([#487](https://github.com/superlinked/sie/issues/487)) ([56d932e](https://github.com/superlinked/sie/commit/56d932ea60364cdf9e3961945324f4b476d71c92))
+* **examples:** reproduce the image retrieval pilot ([#602](https://github.com/superlinked/sie/issues/602)) ([ed79362](https://github.com/superlinked/sie/commit/ed7936255757919b14df3f06a70899c9f6ba1526))
+* **gateway:** bridge a lane its transport reports cold ([#592](https://github.com/superlinked/sie/issues/592)) ([b5d6253](https://github.com/superlinked/sie/commit/b5d6253e7fa2625fdf2a3ba8b96aca0b6a8bd920))
+* **gateway:** bridge buffered cluster generation refusals ([#547](https://github.com/superlinked/sie/issues/547)) ([6ec8dec](https://github.com/superlinked/sie/commit/6ec8dec33f364dbd05d4662424c77cc4b85ea721))
+* **gateway:** bridge cluster extraction and audio refusals ([#549](https://github.com/superlinked/sie/issues/549)) ([b009e9f](https://github.com/superlinked/sie/commit/b009e9f40c62a20a6f3a36ec0e5c5a4db95d1467))
+* **gateway:** bridge encode and score under a numerical admission ([#587](https://github.com/superlinked/sie/issues/587)) ([f6e0820](https://github.com/superlinked/sie/commit/f6e0820715f3f3328fd4db9acfddaa25d6c5cfe4)), closes [#415](https://github.com/superlinked/sie/issues/415)
+* **gateway:** bridge governed generation to the remote route a policy names ([#607](https://github.com/superlinked/sie/issues/607)) ([e875a65](https://github.com/superlinked/sie/commit/e875a65eda3a02d228498b78467445ed9a34b191))
+* **gateway:** carry generation GPU time ([11612e4](https://github.com/superlinked/sie/commit/11612e4e9c6435fc69bc1f55a0c4a49cc5679dca))
+* **gateway:** carry sealed generation GPU time ([2a4a3a1](https://github.com/superlinked/sie/commit/2a4a3a1bb47e7ea58860bf6cfeb0384a73077b09))
+* **gateway:** carry the fallback reason on every bridged work item ([#567](https://github.com/superlinked/sie/issues/567)) ([8823a89](https://github.com/superlinked/sie/commit/8823a895228fd2611e5957e225ef6321d2823ad2))
+* **gateway:** carry the requested model on queued work ([#504](https://github.com/superlinked/sie/issues/504)) ([c98bfbb](https://github.com/superlinked/sie/commit/c98bfbb038dcff59411bcb925d1bf59f70c0ef44))
+* **gateway:** coordinate threshold demand with bounded leases ([#554](https://github.com/superlinked/sie/issues/554)) ([6e1424a](https://github.com/superlinked/sie/commit/6e1424a205514be969b385a88a038af07b86d3c7))
+* **gateway:** disclose which side served on gateway responses ([#502](https://github.com/superlinked/sie/issues/502)) ([04e6159](https://github.com/superlinked/sie/commit/04e6159e97c058cce1d8915b84acd9c3ab95e4c0))
+* **gateway:** enforce remote-forbid with verified worker dispatch ([#546](https://github.com/superlinked/sie/issues/546)) ([19be847](https://github.com/superlinked/sie/commit/19be847b309803ca7c8b7a012670ea946d2ed66b))
+* **gateway:** give a 503 serving refusal its own admission outcome ([#493](https://github.com/superlinked/sie/issues/493)) ([39a1aa9](https://github.com/superlinked/sie/commit/39a1aa95289b35f98b2eff2c07f8dd944b254b23))
+* **gateway:** let a model access policy admit remote routes ([#591](https://github.com/superlinked/sie/issues/591)) ([7e7050f](https://github.com/superlinked/sie/commit/7e7050fd94a73425071f0a65b320dac8c9b3317a))
+* **gateway:** opt in to pre-acceptance remote spill ([#550](https://github.com/superlinked/sie/issues/550)) ([4f3a62e](https://github.com/superlinked/sie/commit/4f3a62efbc7ea05023afa8adf37815447dc9f5d1))
+* **gateway:** preserve validated remote routing policies ([#542](https://github.com/superlinked/sie/issues/542)) ([f05efe5](https://github.com/superlinked/sie/commit/f05efe5d994eef41445e120a58fdaf813a31b1a9))
+* **gateway:** publish load-only model readiness work ([#541](https://github.com/superlinked/sie/issues/541)) ([bd945d6](https://github.com/superlinked/sie/commit/bd945d6c4ce7c4322e239d94b4fbb55a70debe57))
+* **gateway:** restore cluster stream refusals before output ([#548](https://github.com/superlinked/sie/issues/548)) ([0ebc55c](https://github.com/superlinked/sie/commit/0ebc55c087bac5fa1f53b484feec691c415f9911))
+* **helm:** add a remote worker pool that alone holds upstream credentials ([#496](https://github.com/superlinked/sie/issues/496)) ([4747581](https://github.com/superlinked/sie/commit/474758105b9f6f9dc265a7b4c4d7bdc3175e19b4))
+* **helm:** give sie-config the upstream names the remote lanes define ([#508](https://github.com/superlinked/sie/issues/508)) ([72b48d6](https://github.com/superlinked/sie/commit/72b48d6ef15f2d1ba6555681ea75798fdfab4732))
+* **helm:** limit the remote lanes' network access with a NetworkPolicy ([#498](https://github.com/superlinked/sie/issues/498)) ([f81fab3](https://github.com/superlinked/sie/commit/f81fab384a1cfbbb25a7d34ad8a0f3874e8807b3))
+* **helm:** mount equivalence evidence into remote lanes ([#577](https://github.com/superlinked/sie/issues/577)) ([738cf38](https://github.com/superlinked/sie/commit/738cf38d7a3a39bfcb559c1620d1754aa9d82841))
+* **models:** add a 1120 soft-token image profile for Gemma 4 31B on H100 ([#633](https://github.com/superlinked/sie/issues/633)) ([d97dcf2](https://github.com/superlinked/sie/commit/d97dcf24617493198063222d3fdbc55c03abc57d))
+* **models:** add a compact 768-visual-token profile to tomoro-colqwen3-embed-4b ([#469](https://github.com/superlinked/sie/issues/469)) ([92a2a85](https://github.com/superlinked/sie/commit/92a2a85724becb3f1005f5eb5fe913bdb81601e6))
+* **models:** add an 8,192-token output Gemma 4 31B hi-res profile ([#635](https://github.com/superlinked/sie/issues/635)) ([b9231ce](https://github.com/superlinked/sie/commit/b9231ced90b35bf0e813ee0d8b23b3a2a561aa08))
+* **models:** add Qdrant/splade-ecommerce-esci sparse model ([#631](https://github.com/superlinked/sie/issues/631)) ([842c3b0](https://github.com/superlinked/sie/commit/842c3b0aca7c4e533a1ce225f78e888a03d4ecf5))
+* **models:** refresh Qwen3.5-122B candidate ([44047bd](https://github.com/superlinked/sie/commit/44047bd17710ea85ac34dba985a255a28fd80ac3))
+* **models:** refresh Qwen3.5-122B candidate ([e72f50f](https://github.com/superlinked/sie/commit/e72f50f4cb160e96313fd782e7b7f352448a90ea))
+* **queue:** support load-only work and reply to backend fallback refusals ([#525](https://github.com/superlinked/sie/issues/525)) ([1aa45ce](https://github.com/superlinked/sie/commit/1aa45ce2078e0f4bc589fd63be1ff124e8b09720))
+* **remote:** admit exact fresh OpenAI equivalence evidence ([#535](https://github.com/superlinked/sie/issues/535)) ([c6f077f](https://github.com/superlinked/sie/commit/c6f077f58a31a94f72b0e83174adf0b576b00158))
+* **remote:** admit fresh matching SIE profile identity ([#540](https://github.com/superlinked/sie/issues/540)) ([98b55ec](https://github.com/superlinked/sie/commit/98b55ec091fa279c5b5a3079eb9dc74e94e8ded9))
+* **remote:** bind equivalence evidence to execution identity, not one process ([#572](https://github.com/superlinked/sie/issues/572)) ([21f7c0b](https://github.com/superlinked/sie/commit/21f7c0b7fd5071a1ebcf5c7e74735deab081f962)), closes [#415](https://github.com/superlinked/sie/issues/415)
+* **remote:** expose bounded numerical process diagnostics ([#561](https://github.com/superlinked/sie/issues/561)) ([9f966be](https://github.com/superlinked/sie/commit/9f966be23a56554368d5d0370fae7af2a69bf77f))
+* **remote:** expose diagnostic numerical process snapshots ([#557](https://github.com/superlinked/sie/issues/557)) ([6113be1](https://github.com/superlinked/sie/commit/6113be1a631cf8d0956fdd3a924fe3c01e1a8101))
+* **remote:** measure equivalence through a cluster gateway ([#579](https://github.com/superlinked/sie/issues/579)) ([e964598](https://github.com/superlinked/sie/commit/e96459863c4f4a016acc68040c8d84d51481c11b))
+* **remote:** measure the local serving envelope in the equivalence probe ([#588](https://github.com/superlinked/sie/issues/588)) ([43157ba](https://github.com/superlinked/sie/commit/43157baa32af1ebd4f8201ad74897c2e8b4ff219))
+* **remote:** re-verify the numerical admission before an admitted item runs ([#586](https://github.com/superlinked/sie/issues/586)) ([66f5635](https://github.com/superlinked/sie/commit/66f5635cee783fd0546151dc9ec0704c76b571bf)), closes [#415](https://github.com/superlinked/sie/issues/415)
+* **remote:** report remote admissions in worker health ([#578](https://github.com/superlinked/sie/issues/578)) ([a85c04f](https://github.com/superlinked/sie/commit/a85c04f182439be5130d15e3675e77ef3e42dae3))
+* route remote profiles to a dedicated remote worker bundle ([#492](https://github.com/superlinked/sie/issues/492)) ([ab5d4a9](https://github.com/superlinked/sie/commit/ab5d4a9919b1f8a839bf33408313dbf174f53352))
+* **routing:** enable shared threshold routing behind cluster opt-in ([#555](https://github.com/superlinked/sie/issues/555)) ([93d5660](https://github.com/superlinked/sie/commit/93d56609724e23657d948a8f34a018f320dd5ae2))
+* **sdk:** accept origin-confined configured HTTP clients ([#539](https://github.com/superlinked/sie/issues/539)) ([dea0ca1](https://github.com/superlinked/sie/commit/dea0ca1f771a0ba94729f46c5425da160fffbb0f))
+* **sdk:** forbid remote serving and report which side served ([#495](https://github.com/superlinked/sie/issues/495)) ([17ce8b4](https://github.com/superlinked/sie/commit/17ce8b48bab955d351627e1664bcdfa4e9a2bf22))
+* **sdk:** pass a target language to recommend ([#611](https://github.com/superlinked/sie/issues/611)) ([47cf609](https://github.com/superlinked/sie/commit/47cf60918ac3ed24f1d16a599706e243e73ba1fe))
+* **server:** add a routing block to the model config ([#490](https://github.com/superlinked/sie/issues/490)) ([c02785e](https://github.com/superlinked/sie/commit/c02785e6ed703e4424b8621bdff7bbab22fe3728))
+* **server:** add native GLiNER2.5 and Privacy Filter extraction ([#594](https://github.com/superlinked/sie/issues/594)) ([e01ad1e](https://github.com/superlinked/sie/commit/e01ad1e10e551d9f355947fcd9adf978e1a60a25))
+* **server:** add native ZeRank 2 scoring ([#596](https://github.com/superlinked/sie/issues/596)) ([be8a688](https://github.com/superlinked/sie/commit/be8a688374d6c097096c93dee1bea814d653061d))
+* **server:** add optional cuDNN SDPA startup control ([#650](https://github.com/superlinked/sie/issues/650)) ([8770c72](https://github.com/superlinked/sie/commit/8770c726d96ee8c0a39c1a484d41a9fc530cec1d))
+* **server:** add Parakeet-TDT speech-to-text (nvidia/parakeet-tdt-0.6b-v3) ([#608](https://github.com/superlinked/sie/issues/608)) ([7e2a1d3](https://github.com/superlinked/sie/commit/7e2a1d31e48b3cea7560e06b9c40e2eaa3e1aa41))
+* **server:** add Qwen3Guard-Gen-4B and Qwen3Guard-Gen-0.6B guard models ([#448](https://github.com/superlinked/sie/issues/448)) ([471189c](https://github.com/superlinked/sie/commit/471189c55c95160378d04d4254cdbd8ead9969d5))
+* **server:** add SAM 3 open-vocabulary detection ([#614](https://github.com/superlinked/sie/issues/614)) ([7e7d0f5](https://github.com/superlinked/sie/commit/7e7d0f5adee80d8cc0fad1727d7723ba05c813d3))
+* **server:** answer an upstream that cannot serve yet with a retryable 503 ([#494](https://github.com/superlinked/sie/issues/494)) ([ab0217b](https://github.com/superlinked/sie/commit/ab0217b4dbf2ec697eab4f9c29a3d231336737e3))
+* **server:** bridge cold single-node generation before output ([#533](https://github.com/superlinked/sie/issues/533)) ([bab928f](https://github.com/superlinked/sie/commit/bab928fd1ba4465b65cd286c5b8ac24507695e35))
+* **server:** cap the calls to each upstream and stop calling one that keeps failing ([#506](https://github.com/superlinked/sie/issues/506)) ([1a259cc](https://github.com/superlinked/sie/commit/1a259cc8f05f3a81b27c01b57a19165731ae35cb))
+* **server:** collect process-bound numerical fleet evidence ([#556](https://github.com/superlinked/sie/issues/556)) ([7d0f487](https://github.com/superlinked/sie/commit/7d0f48704029ae2df7259a1257ffe3269e85ec99))
+* **server:** count hybrid upstream generation with the model's tokenizer ([#620](https://github.com/superlinked/sie/issues/620)) ([2f9c107](https://github.com/superlinked/sie/commit/2f9c107309e8d1d0362aab88d3ef6f5b1c867e03))
+* **server:** declare openai upstream endpoints and operator request fields ([#499](https://github.com/superlinked/sie/issues/499)) ([2292823](https://github.com/superlinked/sie/commit/2292823f4139bccb4372adade03b83a48211f921))
+* **server:** default gliner-biomed-large-v1.0 to threshold 0.8, chosen on dev data ([#461](https://github.com/superlinked/sie/issues/461)) ([6c13f28](https://github.com/superlinked/sie/commit/6c13f2890f6af3d637770b94d8a86cbc5a3fcac6))
+* **server:** disclose which side served and report routing in /v1/models ([#445](https://github.com/superlinked/sie/issues/445)) ([946b6fb](https://github.com/superlinked/sie/commit/946b6fb3e8f8d5e61d47745f06a4662dd2d382df))
+* **server:** expose conservative immutable profile identity ([#530](https://github.com/superlinked/sie/issues/530)) ([1c62676](https://github.com/superlinked/sie/commit/1c626766a084e5562cbea0e50b0a98d10ef17f5f))
+* **server:** identify flash BGE-M3 profiles with exact library builds ([#575](https://github.com/superlinked/sie/issues/575)) ([de82bad](https://github.com/superlinked/sie/commit/de82badfdc0c701602c909f2caf70c96010b7ffb)), closes [#415](https://github.com/superlinked/sie/issues/415)
+* **server:** identify the BERT, cross-encoder and Nomic flash adapters ([#581](https://github.com/superlinked/sie/issues/581)) ([cd3cb91](https://github.com/superlinked/sie/commit/cd3cb9119438e251adce4fba35b7ed7bb341c9d5))
+* **server:** load a remote profile without waiting for a local load ([#491](https://github.com/superlinked/sie/issues/491)) ([f53883b](https://github.com/superlinked/sie/commit/f53883ba543375748e015770a9bf42bc3a10be43))
+* **server:** measure remote equivalence against local noise ([#531](https://github.com/superlinked/sie/issues/531)) ([e5cc5c6](https://github.com/superlinked/sie/commit/e5cc5c6504eab6b64cb73bb46aed6f02929c2477))
+* **server:** preserve onboarded templates for direct remote chat ([#537](https://github.com/superlinked/sie/issues/537)) ([5581a23](https://github.com/superlinked/sie/commit/5581a234de068cd77d28b32d633bcb7db68dbc66))
+* **server:** run the remote bundle on transformers 5 so hybrid counting loads transformers-5 tokenizers ([#625](https://github.com/superlinked/sie/issues/625)) ([577f3cf](https://github.com/superlinked/sie/commit/577f3cfa295a9ef1706eaabf1102fdd33ddbb4e1))
+* **server:** serve a cold model through its remote profile on a single node ([#503](https://github.com/superlinked/sie/issues/503)) ([3742178](https://github.com/superlinked/sie/commit/374217831db87e94b7cef0d3b54c9ccbdaa43a9c))
+* **server:** serve a remote-backed embedding model through an SIE upstream ([#437](https://github.com/superlinked/sie/issues/437)) ([7a56b3c](https://github.com/superlinked/sie/commit/7a56b3c6bbdeb2408f9a4392a567175acd16c93e))
+* **server:** serve embeddings and rerank through an openai upstream ([#505](https://github.com/superlinked/sie/issues/505)) ([23cc6df](https://github.com/superlinked/sie/commit/23cc6df2eb6f86d23846459d7fcfa1dc4cec773c))
+* **server:** serve native SIE chat through remote profiles ([#523](https://github.com/superlinked/sie/issues/523)) ([d0869f4](https://github.com/superlinked/sie/commit/d0869f443159a9914f009de71bbfeacc964a9309))
+* **server:** serve OpenAI upstream chat and raw generation ([#528](https://github.com/superlinked/sie/issues/528)) ([21e2ab3](https://github.com/superlinked/sie/commit/21e2ab316eb82aa6f9b9bbd13087538395b8632d))
+* **server:** serve remote chat through queued generation ([#529](https://github.com/superlinked/sie/issues/529)) ([a8bbfa4](https://github.com/superlinked/sie/commit/a8bbfa473b940f627374f4aa54810271fcd39454))
+* **server:** serve score, extract and every encode output through an SIE upstream ([#501](https://github.com/superlinked/sie/issues/501)) ([6344850](https://github.com/superlinked/sie/commit/6344850a54ac76fb464b7b2e04ef9c4508245f52))
+* **server:** SMVE sparse multi-vector encoding, with smve profiles for TopK models ([#518](https://github.com/superlinked/sie/issues/518)) ([9eeff51](https://github.com/superlinked/sie/commit/9eeff51a9074db7a1fe28a8fdb8b9759a5a675c3))
+* **server:** stream generation through a remote SIE upstream ([#514](https://github.com/superlinked/sie/issues/514)) ([d3fa3c1](https://github.com/superlinked/sie/commit/d3fa3c16aa8150747da1fa3b2cd0624ba48a3ebd))
+* **server:** support native GLiNER2 entity descriptions ([#605](https://github.com/superlinked/sie/issues/605)) ([461f8dc](https://github.com/superlinked/sie/commit/461f8dc5131b885d72d272406b78abd5d3f26fc4))
+* **server:** support native LightOnOCR-3 extraction ([#632](https://github.com/superlinked/sie/issues/632)) ([5e60012](https://github.com/superlinked/sie/commit/5e6001278942d0c65420a0afe53a936947bc4eb5))
+* **server:** validate shared remote chat responses ([#513](https://github.com/superlinked/sie/issues/513)) ([42897e9](https://github.com/superlinked/sie/commit/42897e9b4994f86add30173bffe43f0f01b26726))
+* **telemetry:** observe cluster remote fallback persistence ([#551](https://github.com/superlinked/sie/issues/551)) ([8e2032b](https://github.com/superlinked/sie/commit/8e2032b3e7018a6926c5a32a38f2ebd3d3e870aa))
+* **worker:** add versioned execution authority IPC methods ([#544](https://github.com/superlinked/sie/issues/544)) ([5f832a4](https://github.com/superlinked/sie/commit/5f832a4fab9764b5902a5efa5840670c1c42a0b5))
+* **worker:** fence verified dispatch with versioned queues ([#545](https://github.com/superlinked/sie/issues/545)) ([2ba3d6e](https://github.com/superlinked/sie/commit/2ba3d6e3d02894cf4585c6bfb83b8513a1a8c928))
+
+
+### Bug Fixes
+
+* answer /v1/rerank without usage when the score has none ([#569](https://github.com/superlinked/sie/issues/569)) ([3459946](https://github.com/superlinked/sie/commit/3459946f282c61b73f6b0e1e1f5a0b25a317121b))
+* **batcher:** serve long-form audio in its own lane, alternating with clips ([#601](https://github.com/superlinked/sie/issues/601)) ([45b6ffe](https://github.com/superlinked/sie/commit/45b6ffe1b58dd660a17ad0e2e5f9f9f174935b68)), closes [#585](https://github.com/superlinked/sie/issues/585)
+* **examples:** align graph evidence with recorded calls ([#510](https://github.com/superlinked/sie/issues/510)) ([ef90dd6](https://github.com/superlinked/sie/commit/ef90dd602978cd7b02fe8820473e83e0f0f88191))
+* **examples:** image-search scores SigLIP so400m-384, the model superlinked.com/image-search sells ([#476](https://github.com/superlinked/sie/issues/476)) ([395366c](https://github.com/superlinked/sie/commit/395366cd58d0c14eddfd59c84daec438acb5d8c8))
+* **examples:** keep journal replay outside latency timing ([10564f8](https://github.com/superlinked/sie/commit/10564f8896407b390f9323f82ddc60160d1b99f2))
+* **examples:** redact credentials in reply keys ([b57c033](https://github.com/superlinked/sie/commit/b57c0333b4b5e2068c6bf1dfbf81cba533ffd8ad))
+* **examples:** reject malformed latency reply members ([e068b1d](https://github.com/superlinked/sie/commit/e068b1d8fa779c91fe69776389b4dd95415b8b95))
+* **examples:** rerun and grade document-field extraction with the measured recipe ([#622](https://github.com/superlinked/sie/issues/622)) ([557f856](https://github.com/superlinked/sie/commit/557f8560d702c9ee00fddfa4ba2291cc91c106c0))
+* **examples:** retain owned replies in timed event capture ([ab247dc](https://github.com/superlinked/sie/commit/ab247dc2979b19317e934d08a543e78a13228b4a))
+* **examples:** scope OCR quality claims and support self-hosting ([#483](https://github.com/superlinked/sie/issues/483)) ([6cccc04](https://github.com/superlinked/sie/commit/6cccc04de9c6b78d3217854ad2377de36c0ffbd2))
+* **examples:** sync latency outputs and retain deadline reasons ([c4a9274](https://github.com/superlinked/sie/commit/c4a927498abab822acb947f586d1187769374f89))
+* **examples:** validate latency trial execution evidence ([c4e94a3](https://github.com/superlinked/sie/commit/c4e94a3639673b19bf991658ec81bf80a2b7da4b))
+* **gateway:** answer a queued generation INPUT_TOO_LONG with 400 ([#497](https://github.com/superlinked/sie/issues/497)) ([0cb086d](https://github.com/superlinked/sie/commit/0cb086ddca0427b935524bc6a692fd8cbd3b9bff))
+* **gateway:** derive a failed bridge's fallback error as the single server does ([#580](https://github.com/superlinked/sie/issues/580)) ([70b3181](https://github.com/superlinked/sie/commit/70b318162b452c48b6cc3f805e9dc3113386ce44))
+* **gateway:** dispatch a forbid request normally when its model has no remote route ([#564](https://github.com/superlinked/sie/issues/564)) ([2600ded](https://github.com/superlinked/sie/commit/2600ded6831918106bdaa94936b243782c81be04))
+* **gateway:** keep a hidden remote profile out of the model listing ([#598](https://github.com/superlinked/sie/issues/598)) ([7a43250](https://github.com/superlinked/sie/commit/7a43250ddc1a5541fe1024a2ab11063979273f66))
+* **gateway:** keep json_schema property order through to the worker ([#455](https://github.com/superlinked/sie/issues/455)) ([571f45b](https://github.com/superlinked/sie/commit/571f45bf5c49caf85d7780c4b224c2c5b5b33173))
+* **gateway:** keep requests for undeclared outputs off numerical bridges ([#597](https://github.com/superlinked/sie/issues/597)) ([b6f1702](https://github.com/superlinked/sie/commit/b6f17028156b30096cade2bb36bdd8455ff681f1))
+* **helm:** require pytorch for remote lanes ([2e1aef2](https://github.com/superlinked/sie/commit/2e1aef270914c849c0c93221286a0afce1bf61a9))
+* **helm:** require pytorch for remote lanes ([1545d61](https://github.com/superlinked/sie/commit/1545d6118ce8a1fd2226a6efed7d358fb1c97bc8))
+* match identifier patterns against the whole value ([#568](https://github.com/superlinked/sie/issues/568)) ([6e5ce3d](https://github.com/superlinked/sie/commit/6e5ce3d5dc6e11bfd48f06d11bf16be8d72c3ce2))
+* **models:** pin GLiNER v2.5 tokenizer dependencies ([#634](https://github.com/superlinked/sie/issues/634)) ([a52cdbb](https://github.com/superlinked/sie/commit/a52cdbb0b5bd2a14042b441df6e6e5e867f2c944))
+* **models:** raise the Hy-MT2-1.8B output cap to 1024 tokens ([#610](https://github.com/superlinked/sie/issues/610)) ([e831d4d](https://github.com/superlinked/sie/commit/e831d4d89d733453168f951208472f872c731a99))
+* **models:** validate Qwen runtime defaults ([8fae464](https://github.com/superlinked/sie/commit/8fae464d16cdc138efffcd0f1a22e201a2bffa7b))
+* **parakeet:** decode long audio in pieces cut at pauses ([#624](https://github.com/superlinked/sie/issues/624)) ([4d0e3c6](https://github.com/superlinked/sie/commit/4d0e3c6a0680f73db99fea1a9769562bbd367e6f))
+* **queue:** answer a remote profile's upstream refusal at once with its wait ([#570](https://github.com/superlinked/sie/issues/570)) ([22aa825](https://github.com/superlinked/sie/commit/22aa8250d41a67cbb14077e27f75a1ce1dc3b57a))
+* **remote:** bind profile identity to observed hardware and BLAS ([#536](https://github.com/superlinked/sie/issues/536)) ([89640b0](https://github.com/superlinked/sie/commit/89640b0f6ae2c9e90d96d1baf7462b60726e5121))
+* **remote:** bound configuration waits and settle grammar refusals ([#558](https://github.com/superlinked/sie/issues/558)) ([b6bbf38](https://github.com/superlinked/sie/commit/b6bbf38d1e539e470430d573e71d16309bd0eabb))
+* **remote:** carry admitted numerical work on its own subject ([#589](https://github.com/superlinked/sie/issues/589)) ([6d4d74b](https://github.com/superlinked/sie/commit/6d4d74b13fcbaa25df39a0df4e2895a51740517f))
+* **remote:** keep instructions and unadmitted plans off numerical bridges ([#600](https://github.com/superlinked/sie/issues/600)) ([ab016b1](https://github.com/superlinked/sie/commit/ab016b13b4417baabf6964c745eedfda3b52ad0f))
+* **remote:** name both identities when SIE identity admission refuses a mismatch ([#563](https://github.com/superlinked/sie/issues/563)) ([52c6c6d](https://github.com/superlinked/sie/commit/52c6c6d8d7843a412be9fc005f9f6b7ce9d2c50d))
+* **sdk:** preserve HTTP retry hints in terminal server errors ([#560](https://github.com/superlinked/sie/issues/560)) ([7264193](https://github.com/superlinked/sie/commit/72641933c2d822ec2f38ea4c398bc08eeb3e3be3))
+* **sdk:** serialize encode query roles in nested options ([#617](https://github.com/superlinked/sie/issues/617)) ([812962d](https://github.com/superlinked/sie/commit/812962d5a6ad22e7d5f08f75b1ab719789503ce4))
+* **server:** apply default_instruction in the remaining flash embedders ([#623](https://github.com/superlinked/sie/issues/623)) ([07983db](https://github.com/superlinked/sie/commit/07983dbfe60812533468989318f7bcbaf726f5d1))
+* **server:** apply default_instruction to qwen2_flash queries ([#621](https://github.com/superlinked/sie/issues/621)) ([db4c0a8](https://github.com/superlinked/sie/commit/db4c0a86d90e30b21a008dadfe9cd52fcbd043aa))
+* **server:** classify all of a long text with GLiNER2, not only its first 512 words ([#449](https://github.com/superlinked/sie/issues/449)) ([82cbf2e](https://github.com/superlinked/sie/commit/82cbf2ec7f5c757e18e95eba50595c4d5f5806ef))
+* **server:** convert queued tool-call arguments by the declared schema ([#609](https://github.com/superlinked/sie/issues/609)) ([5f00470](https://github.com/superlinked/sie/commit/5f0047091c907a92e43e3bfd764fd3d455c5793f))
+* **server:** default grammar-constrained generation to greedy sampling ([#453](https://github.com/superlinked/sie/issues/453)) ([b1c8887](https://github.com/superlinked/sie/commit/b1c88878833fbb0367660302520e71657fbdfcd5))
+* **server:** exit on SIGTERM after starting the drain ([#637](https://github.com/superlinked/sie/issues/637)) ([ecb957d](https://github.com/superlinked/sie/commit/ecb957db6f4503ac7e777be74c4b3a9a50e8a260))
+* **server:** give background identity refreshes their own budget ([#618](https://github.com/superlinked/sie/issues/618)) ([c998453](https://github.com/superlinked/sie/commit/c99845360eab0f7f11dbe5a67966dda3de081ecf))
+* **server:** honor explicitly selected default profiles ([#534](https://github.com/superlinked/sie/issues/534)) ([87a0759](https://github.com/superlinked/sie/commit/87a0759434d97a6f4316916652e49e71132c150b))
+* **server:** ignore optional triton type imports ([10cac53](https://github.com/superlinked/sie/commit/10cac53729f760f0619f01cf440cbaf74e8fbab7))
+* **server:** ignore optional triton type imports ([f9ab207](https://github.com/superlinked/sie/commit/f9ab2079389ba3f6f8df63e5714d7abfbb3a3e6f))
+* **server:** let GLiNER checkpoints on byte-level BPE encoders read split words ([#460](https://github.com/superlinked/sie/issues/460)) ([d214552](https://github.com/superlinked/sie/commit/d214552a699dfd591ead7557e48157ea0bd42253))
+* **server:** meter ColQwen3 text from processed inputs ([#509](https://github.com/superlinked/sie/issues/509)) ([d787725](https://github.com/superlinked/sie/commit/d787725f825b451e650799c8fb8c515fea46bbd2))
+* **server:** pin GLiFormer decoding probe to float32 ([#562](https://github.com/superlinked/sie/issues/562)) ([df09d4b](https://github.com/superlinked/sie/commit/df09d4b7c94de438e58cc44fa0e325e12cac2288))
+* **server:** read long documents whole in GLiNER extract ([#451](https://github.com/superlinked/sie/issues/451)) ([2f97d02](https://github.com/superlinked/sie/commit/2f97d022440f61403efea9badc97c55a16bb45ce))
+* **server:** refuse LoRAs that train biases ([#583](https://github.com/superlinked/sie/issues/583)) ([dcfdf51](https://github.com/superlinked/sie/commit/dcfdf51e695be990f0abed5b006c8ec47ab544b9))
+* **server:** refuse remote generation before streaming headers ([#532](https://github.com/superlinked/sie/issues/532)) ([9719ae1](https://github.com/superlinked/sie/commit/9719ae16413d59117de55bdd72d125f5ea7d0153))
+* **server:** reject a config delta per model instead of stalling its bundle ([#489](https://github.com/superlinked/sie/issues/489)) ([81cc427](https://github.com/superlinked/sie/commit/81cc42789544bbc3700769771b7d409c4228b846))
+* **server:** reject incomplete GLiNER2 extraction items ([#552](https://github.com/superlinked/sie/issues/552)) ([689e4d8](https://github.com/superlinked/sie/commit/689e4d808cbd66008500da8f58d8fa6e71510b8f))
+* **server:** report GLiNER2 caller errors as invalid input ([#526](https://github.com/superlinked/sie/issues/526)) ([96066a3](https://github.com/superlinked/sie/commit/96066a311fd991ed32b57ece62fb92b09ae47ea1))
+* **server:** report NLI zero-shot extract usage and enforce the extract label limit on the queue path ([#447](https://github.com/superlinked/sie/issues/447)) ([2cf3d89](https://github.com/superlinked/sie/commit/2cf3d89382955f1b2cf592662de93cb332303f47))
+* **server:** score a lone GLiClass label on its own in single-label mode ([#616](https://github.com/superlinked/sie/issues/616)) ([45c3fee](https://github.com/superlinked/sie/commit/45c3fee8a4d649e987e93b0c899f947e10393b61))
+* **server:** score NLI zero-shot labels like the transformers pipeline ([#615](https://github.com/superlinked/sie/issues/615)) ([d134ab2](https://github.com/superlinked/sie/commit/d134ab29815d092d7773b375a9cc5d7be180a3a0))
+* **server:** score Qwen3 text rerankers in float32 so top candidates stop tying ([#440](https://github.com/superlinked/sie/issues/440)) ([8de63e3](https://github.com/superlinked/sie/commit/8de63e3f3e7719b4152fe7babfb8a1733277fd4f))
+* **server:** size NLI zero-shot extract batches by the rows each item runs ([#465](https://github.com/superlinked/sie/issues/465)) ([57a9856](https://github.com/superlinked/sie/commit/57a98562cba9d1b9f54fab79f14e5654f35a2402))
+* **server:** skip remote profiles a worker refuses in pool isolation ([#627](https://github.com/superlinked/sie/issues/627)) ([cb79fc1](https://github.com/superlinked/sie/commit/cb79fc1eafff8d7ffc51d6ade190b05c4422c0ef))
+* **server:** suppress private chat reasoning logprobs independently of thinking policy ([#527](https://github.com/superlinked/sie/issues/527)) ([d9d63dc](https://github.com/superlinked/sie/commit/d9d63dcca3c900cb7a49bd832eca1c287587a5bf))
+* **sglang:** bound JSON number digits in XGrammar json_schema grammars ([#467](https://github.com/superlinked/sie/issues/467)) ([7f3666d](https://github.com/superlinked/sie/commit/7f3666dd785a215274eb2569f3a9aee9a4826a00))
+* validate remote routing before configuration writes ([#512](https://github.com/superlinked/sie/issues/512)) ([cb95d4a](https://github.com/superlinked/sie/commit/cb95d4ac753492d5d854cdcc6a61ab995e207948))
+* **whisper:** transcribe long-form audio one recording per pipeline call ([#582](https://github.com/superlinked/sie/issues/582)) ([8bdaeba](https://github.com/superlinked/sie/commit/8bdaebac7e7480c19d5254e8b5161f5293e9a745)), closes [#574](https://github.com/superlinked/sie/issues/574)
+* **worker:** enforce live configuration authority through execution ([#543](https://github.com/superlinked/sie/issues/543)) ([f6c530c](https://github.com/superlinked/sie/commit/f6c530cf9cce813d052deca316b4a1e439514f77))
+
+
+### Performance Improvements
+
+* **colbert:** tokenize a ModernColBERT batch in one call and pack it on the host ([#646](https://github.com/superlinked/sie/issues/646)) ([e15d014](https://github.com/superlinked/sie/commit/e15d0149d611e1a5f5f1d877ddac57ebecbbcddf))
+* **gliner:** run 32 rows per forward pass instead of gliner's default 8 ([#638](https://github.com/superlinked/sie/issues/638)) ([e7622c3](https://github.com/superlinked/sie/commit/e7622c3d314e8c104fc73dccf40c728ab9ac9718))
+* **gliner:** turn the adaptive batching controller off on the GLiNER profiles ([#648](https://github.com/superlinked/sie/issues/648)) ([cfaafe3](https://github.com/superlinked/sie/commit/cfaafe356bca2f7c320e53aa834651e48bea04f8))
+* **models:** add a 16-request CUDA-graph H100 profile for Qwen3.8-27B-FP8 ([#454](https://github.com/superlinked/sie/issues/454)) ([409fee6](https://github.com/superlinked/sie/commit/409fee6709ed1b55cc9cdcfa886d6c7a37a88cac))
+* **models:** compile compact JSON grammars on every Qwen3.6-35B-A3B profile ([#636](https://github.com/superlinked/sie/issues/636)) ([d8f58cb](https://github.com/superlinked/sie/commit/d8f58cb85dbe69564b2ba0ebf8718dd7dd955eba))
+* **models:** compile compact, digit-bounded JSON grammars on the Gemma 4 31B hi-res profiles ([#641](https://github.com/superlinked/sie/issues/641)) ([a8eb93d](https://github.com/superlinked/sie/commit/a8eb93d9d7ee2a910f3256e6486454d897a81f24))
+* **models:** enable decode CUDA graphs on the Qwen3.8-27B-FP8 bare route and overlap the batch lane ([#640](https://github.com/superlinked/sie/issues/640)) ([0f0d691](https://github.com/superlinked/sie/commit/0f0d691beae91d6177b660e3b6d68f67f7383556))
+* **models:** read a single Qwen3.8-27B-FP8 page image at 3.2 megapixels and compile compact JSON grammars ([#466](https://github.com/superlinked/sie/issues/466)) ([329df87](https://github.com/superlinked/sie/commit/329df878acd18ffdd5ca3decba5605a619c16840))
+* **models:** serve Qwen3.6-35B-A3B default on the CUDA 13 engine with CUDA graphs ([#639](https://github.com/superlinked/sie/issues/639)) ([2f7fcbd](https://github.com/superlinked/sie/commit/2f7fcbda0e2d86703bcfe5a844d580e7992ee571))
+* **qwen3-embedding:** fuse the 8B's elementwise layer work into exact Triton kernels ([#645](https://github.com/superlinked/sie/issues/645)) ([6949a8b](https://github.com/superlinked/sie/commit/6949a8b935c1f29950b4dd6b0f333cc224298e54))
+* **qwen3-embedding:** keep up to four Qwen3-Embedding-4B batches in flight to SGLang ([#647](https://github.com/superlinked/sie/issues/647)) ([99140a4](https://github.com/superlinked/sie/commit/99140a475308de3199e3aba8302b3fcfa7fa3b00))
+* **sam3:** skip the unused mask decoder, encode images one at a time and cache label encodings ([#643](https://github.com/superlinked/sie/issues/643)) ([06bc893](https://github.com/superlinked/sie/commit/06bc893a256775d694cbcc4c38cb54fc20f1ac40))
+* **server:** run SGLang OCR batches concurrently instead of one at a time ([#462](https://github.com/superlinked/sie/issues/462)) ([dce7611](https://github.com/superlinked/sie/commit/dce7611654a3bd25b76405dbdb0415c8ec92cd40))
+* **server:** serve OWLv2 with the fast image processor and shrink oversized photos ([#473](https://github.com/superlinked/sie/issues/473)) ([b01062d](https://github.com/superlinked/sie/commit/b01062d8c3566dfb27d6e00ce9291d713a842d43))
+* **sparse:** pool before the sparse activation and keep doc-v3 batches at full size ([#642](https://github.com/superlinked/sie/issues/642)) ([9075c1f](https://github.com/superlinked/sie/commit/9075c1f6373f0f728ebb205b4f94d339ec8dfb09))
+
 ## [0.9.0](https://github.com/superlinked/sie/compare/v0.8.3...v0.9.0) (2026-09-30)
 
 
diff --git a/Cargo.lock b/Cargo.lock
index f2ee3e7e4..856102f68 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -3774,7 +3774,7 @@ checksum = "f8fadd59c855ef2080decdef8ff161eb6661b86933c9d82e5ba29dc602a55aba"
 
 [[package]]
 name = "sie-audio-prep"
-version = "0.9.0"
+version = "0.10.0"
 dependencies = [
  "audioadapter-buffers",
  "opusic-sys",
@@ -3788,7 +3788,7 @@ dependencies = [
 
 [[package]]
 name = "sie-gateway"
-version = "0.9.0"
+version = "0.10.0"
 dependencies = [
  "arc-swap",
  "async-nats",
@@ -3850,7 +3850,7 @@ dependencies = [
 
 [[package]]
 name = "sie-server-sidecar"
-version = "0.9.0"
+version = "0.10.0"
 dependencies = [
  "anyhow",
  "async-nats",
diff --git a/deploy/helm/sie-cluster/Chart.yaml b/deploy/helm/sie-cluster/Chart.yaml
index 71d08fc18..92caae10c 100644
--- a/deploy/helm/sie-cluster/Chart.yaml
+++ b/deploy/helm/sie-cluster/Chart.yaml
@@ -4,8 +4,8 @@ description: >-
   SIE (Search Inference Engine) Kubernetes deployment with gateway, GPU workers,
   and observability
 type: application
-version: 0.9.0  # x-release-please-version
-appVersion: v0.9.0  # x-release-please-version
+version: 0.10.0  # x-release-please-version
+appVersion: v0.10.0  # x-release-please-version
 keywords:
   - inference
   - embedding
diff --git a/integrations/sie_chroma/pyproject.toml b/integrations/sie_chroma/pyproject.toml
index 8ead31e8e..71cbc1f81 100644
--- a/integrations/sie_chroma/pyproject.toml
+++ b/integrations/sie_chroma/pyproject.toml
@@ -1,6 +1,6 @@
 [project]
 name = "sie-chroma"
-version = "0.9.0"
+version = "0.10.0"
 description = "SIE integration for ChromaDB"
 readme = "README.md"
 requires-python = ">=3.10"
diff --git a/integrations/sie_crewai/pyproject.toml b/integrations/sie_crewai/pyproject.toml
index 4ca8fc6ea..b7a9594be 100644
--- a/integrations/sie_crewai/pyproject.toml
+++ b/integrations/sie_crewai/pyproject.toml
@@ -1,6 +1,6 @@
 [project]
 name = "sie-crewai"
-version = "0.9.0"
+version = "0.10.0"
 description = "SIE integration for CrewAI"
 readme = "README.md"
 requires-python = ">=3.10"
diff --git a/integrations/sie_dspy/pyproject.toml b/integrations/sie_dspy/pyproject.toml
index 35d577018..93e23347e 100644
--- a/integrations/sie_dspy/pyproject.toml
+++ b/integrations/sie_dspy/pyproject.toml
@@ -1,6 +1,6 @@
 [project]
 name = "sie-dspy"
-version = "0.9.0"
+version = "0.10.0"
 description = "SIE integration for DSPy"
 readme = "README.md"
 requires-python = ">=3.10"
diff --git a/integrations/sie_haystack/pyproject.toml b/integrations/sie_haystack/pyproject.toml
index 181b4698d..a317cd1a7 100644
--- a/integrations/sie_haystack/pyproject.toml
+++ b/integrations/sie_haystack/pyproject.toml
@@ -1,6 +1,6 @@
 [project]
 name = "sie-haystack"
-version = "0.9.0"
+version = "0.10.0"
 description = "SIE integration for Haystack"
 readme = "README.md"
 requires-python = ">=3.10"
diff --git a/integrations/sie_lancedb/pyproject.toml b/integrations/sie_lancedb/pyproject.toml
index 003c7b8bd..608438382 100644
--- a/integrations/sie_lancedb/pyproject.toml
+++ b/integrations/sie_lancedb/pyproject.toml
@@ -1,6 +1,6 @@
 [project]
 name = "sie-lancedb"
-version = "0.9.0"
+version = "0.10.0"
 description = "SIE integration for LanceDB"
 requires-python = ">=3.12"
 dependencies = [
diff --git a/integrations/sie_langchain/pyproject.toml b/integrations/sie_langchain/pyproject.toml
index 523f148ca..0b8efeff1 100644
--- a/integrations/sie_langchain/pyproject.toml
+++ b/integrations/sie_langchain/pyproject.toml
@@ -1,6 +1,6 @@
 [project]
 name = "sie-langchain"
-version = "0.9.0"
+version = "0.10.0"
 description = "SIE integration for LangChain"
 requires-python = ">=3.10"
 dependencies = [
diff --git a/integrations/sie_llamaindex/pyproject.toml b/integrations/sie_llamaindex/pyproject.toml
index 01343dea6..d1243b44a 100644
--- a/integrations/sie_llamaindex/pyproject.toml
+++ b/integrations/sie_llamaindex/pyproject.toml
@@ -1,6 +1,6 @@
 [project]
 name = "sie-llamaindex"
-version = "0.9.0"
+version = "0.10.0"
 description = "SIE integration for LlamaIndex"
 requires-python = ">=3.10"
 dependencies = [
diff --git a/integrations/sie_qdrant/pyproject.toml b/integrations/sie_qdrant/pyproject.toml
index e5d678c79..573af55d4 100644
--- a/integrations/sie_qdrant/pyproject.toml
+++ b/integrations/sie_qdrant/pyproject.toml
@@ -1,6 +1,6 @@
 [project]
 name = "sie-qdrant"
-version = "0.9.0"
+version = "0.10.0"
 description = "SIE integration for Qdrant"
 readme = "README.md"
 requires-python = ">=3.10"
diff --git a/integrations/sie_ts_chroma/package.json b/integrations/sie_ts_chroma/package.json
index dbded145a..c4f060ac1 100644
--- a/integrations/sie_ts_chroma/package.json
+++ b/integrations/sie_ts_chroma/package.json
@@ -1,6 +1,6 @@
 {
   "name": "@superlinked/sie-chroma",
-  "version": "0.9.0",
+  "version": "0.10.0",
   "description": "SIE embedding function for ChromaDB",
   "type": "module",
   "main": "./dist/index.cjs",
diff --git a/integrations/sie_ts_lancedb/package.json b/integrations/sie_ts_lancedb/package.json
index 5de1e93d5..b4cf55c64 100644
--- a/integrations/sie_ts_lancedb/package.json
+++ b/integrations/sie_ts_lancedb/package.json
@@ -1,6 +1,6 @@
 {
   "name": "@superlinked/sie-lancedb",
-  "version": "0.9.0",
+  "version": "0.10.0",
   "description": "SIE embedding function and reranker for LanceDB",
   "type": "module",
   "main": "./dist/index.cjs",
diff --git a/integrations/sie_ts_langchain/package.json b/integrations/sie_ts_langchain/package.json
index df0dba97c..09fe11adf 100644
--- a/integrations/sie_ts_langchain/package.json
+++ b/integrations/sie_ts_langchain/package.json
@@ -1,6 +1,6 @@
 {
   "name": "@superlinked/sie-langchain",
-  "version": "0.9.0",
+  "version": "0.10.0",
   "description": "SIE embeddings integration for LangChain.js",
   "type": "module",
   "main": "./dist/index.cjs",
diff --git a/integrations/sie_ts_llamaindex/package.json b/integrations/sie_ts_llamaindex/package.json
index 2db86bd48..fb07d5f83 100644
--- a/integrations/sie_ts_llamaindex/package.json
+++ b/integrations/sie_ts_llamaindex/package.json
@@ -1,6 +1,6 @@
 {
   "name": "@superlinked/sie-llamaindex",
-  "version": "0.9.0",
+  "version": "0.10.0",
   "description": "SIE embeddings integration for LlamaIndex.TS",
   "type": "module",
   "main": "./dist/index.cjs",
diff --git a/integrations/sie_weaviate/pyproject.toml b/integrations/sie_weaviate/pyproject.toml
index eb1c89551..fb617adb3 100644
--- a/integrations/sie_weaviate/pyproject.toml
+++ b/integrations/sie_weaviate/pyproject.toml
@@ -1,6 +1,6 @@
 [project]
 name = "sie-weaviate"
-version = "0.9.0"
+version = "0.10.0"
 description = "SIE integration for Weaviate"
 readme = "README.md"
 requires-python = ">=3.10"
diff --git a/packages/sie_audio_prep/Cargo.toml b/packages/sie_audio_prep/Cargo.toml
index dbfb81ab2..bf68b8229 100644
--- a/packages/sie_audio_prep/Cargo.toml
+++ b/packages/sie_audio_prep/Cargo.toml
@@ -1,6 +1,6 @@
 [package]
 name = "sie-audio-prep"
-version = "0.9.0"
+version = "0.10.0"
 edition = "2021"
 license = "Apache-2.0"
 description = "Bounded, shared audio decoding and resampling for SIE workers."
diff --git a/packages/sie_audio_prep/build_wheel.py b/packages/sie_audio_prep/build_wheel.py
index ee6b45bf9..755b98cad 100644
--- a/packages/sie_audio_prep/build_wheel.py
+++ b/packages/sie_audio_prep/build_wheel.py
@@ -16,7 +16,7 @@
 from pathlib import Path
 
 MATURIN_VERSION = "1.13.3"
-AUDIO_PREP_VERSION = "0.9.0"  # x-release-please-version
+AUDIO_PREP_VERSION = "0.10.0"  # x-release-please-version
 AUDIO_WHEEL_COMPATIBILITY = "manylinux_2_28"
 AUDIO_WHEEL_FILENAME = f"sie_audio_prep-{AUDIO_PREP_VERSION}-cp312-abi3-{AUDIO_WHEEL_COMPATIBILITY}_x86_64.whl"
 AUDIO_WHEEL_REMOTE = f"/opt/sie/wheels/{AUDIO_WHEEL_FILENAME}"
diff --git a/packages/sie_audio_prep/pyproject.toml b/packages/sie_audio_prep/pyproject.toml
index c9feeac5d..5c9bbbdce 100644
--- a/packages/sie_audio_prep/pyproject.toml
+++ b/packages/sie_audio_prep/pyproject.toml
@@ -4,7 +4,7 @@ build-backend = "maturin"
 
 [project]
 name = "sie-audio-prep"
-version = "0.9.0"
+version = "0.10.0"
 description = "Bounded, shared audio decoding and resampling for SIE workers."
 requires-python = ">=3.12,<3.13"
 license = { text = "Apache-2.0" }
diff --git a/packages/sie_config/pyproject.toml b/packages/sie_config/pyproject.toml
index 8f8980ebf..f86cc3d8b 100644
--- a/packages/sie_config/pyproject.toml
+++ b/packages/sie_config/pyproject.toml
@@ -1,6 +1,6 @@
 [project]
 name = "sie-config"
-version = "0.9.0"
+version = "0.10.0"
 description = "SIE Config Service - Config control plane for SIE clusters"
 requires-python = ">=3.12,<3.13"
 license = { text = "Apache-2.0" }
diff --git a/packages/sie_gateway/Cargo.toml b/packages/sie_gateway/Cargo.toml
index e84cd0906..5fd725043 100644
--- a/packages/sie_gateway/Cargo.toml
+++ b/packages/sie_gateway/Cargo.toml
@@ -1,6 +1,6 @@
 [package]
 name = "sie-gateway"
-version = "0.9.0"
+version = "0.10.0"
 edition = "2021"
 publish = false
 
diff --git a/packages/sie_gateway/openapi.json b/packages/sie_gateway/openapi.json
index 7669b6241..b1a99f58a 100644
--- a/packages/sie_gateway/openapi.json
+++ b/packages/sie_gateway/openapi.json
@@ -4246,7 +4246,7 @@
       "url": "https://www.apache.org/licenses/LICENSE-2.0.html"
     },
     "title": "SIE Gateway",
-    "version": "0.9.0"
+    "version": "0.10.0"
   },
   "openapi": "3.1.0",
   "paths": {
diff --git a/packages/sie_mcp/pyproject.toml b/packages/sie_mcp/pyproject.toml
index 62dd3ef9f..f791190ca 100644
--- a/packages/sie_mcp/pyproject.toml
+++ b/packages/sie_mcp/pyproject.toml
@@ -1,6 +1,6 @@
 [project]
 name = "sie-mcp"
-version = "0.9.0"
+version = "0.10.0"
 description = "SIE MCP edge - hosted MCP server exposing SIE document jobs"
 requires-python = ">=3.12,<3.13"
 license = { text = "Apache-2.0" }
diff --git a/packages/sie_sdk/pyproject.toml b/packages/sie_sdk/pyproject.toml
index 8c4524b34..fdd7538fb 100644
--- a/packages/sie_sdk/pyproject.toml
+++ b/packages/sie_sdk/pyproject.toml
@@ -1,6 +1,6 @@
 [project]
 name = "sie-sdk"
-version = "0.9.0"
+version = "0.10.0"
 description = "Search Inference Engine - Python SDK"
 requires-python = ">=3.12"
 license = { text = "Apache-2.0" }
diff --git a/packages/sie_server/openapi.json b/packages/sie_server/openapi.json
index 47620c1e0..30c142096 100644
--- a/packages/sie_server/openapi.json
+++ b/packages/sie_server/openapi.json
@@ -3,7 +3,7 @@
   "info": {
     "title": "SIE Server",
     "description": "Search Inference Engine - GPU inference server for search workloads",
-    "version": "0.9.0"
+    "version": "0.10.0"
   },
   "paths": {
     "/healthz": {
diff --git a/packages/sie_server/pyproject.toml b/packages/sie_server/pyproject.toml
index eea29cf5c..5d3b92394 100644
--- a/packages/sie_server/pyproject.toml
+++ b/packages/sie_server/pyproject.toml
@@ -1,6 +1,6 @@
 [project]
 name = "sie-server"
-version = "0.9.0"
+version = "0.10.0"
 description = "Search Inference Engine - GPU inference server for search workloads"
 requires-python = ">=3.12,<3.13"
 license = { text = "Apache-2.0" }
diff --git a/packages/sie_server_sidecar/Cargo.toml b/packages/sie_server_sidecar/Cargo.toml
index eef1bbd4a..784da22ec 100644
--- a/packages/sie_server_sidecar/Cargo.toml
+++ b/packages/sie_server_sidecar/Cargo.toml
@@ -1,6 +1,6 @@
 [package]
 name = "sie-server-sidecar"
-version = "0.9.0"
+version = "0.10.0"
 edition = "2021"
 publish = false
 description = "SIE worker sidecar: NATS JetStream consumer, scheduler, IPC client, and GPU-independent prep for sie-server pods."
diff --git a/packages/sie_ts_sdk/package.json b/packages/sie_ts_sdk/package.json
index c1efa4adf..cd7608bf6 100644
--- a/packages/sie_ts_sdk/package.json
+++ b/packages/sie_ts_sdk/package.json
@@ -1,6 +1,6 @@
 {
   "name": "@superlinked/sie-sdk",
-  "version": "0.9.0",
+  "version": "0.10.0",
   "description": "Official TypeScript SDK for SIE (Search Inference Engine)",
   "type": "module",
   "main": "./dist/index.cjs",
diff --git a/packages/sie_ts_sdk/src/version.ts b/packages/sie_ts_sdk/src/version.ts
index 8dd62d1aa..436ffc6e3 100644
--- a/packages/sie_ts_sdk/src/version.ts
+++ b/packages/sie_ts_sdk/src/version.ts
@@ -1 +1 @@
-export const SDK_VERSION = "0.9.0"; // x-release-please-version
+export const SDK_VERSION = "0.10.0"; // x-release-please-version
diff --git a/uv.lock b/uv.lock
index 11d586fa3..843d66bff 100644
--- a/uv.lock
+++ b/uv.lock
@@ -4476,12 +4476,12 @@ dev = [
 
 [[package]]
 name = "sie-audio-prep"
-version = "0.9.0"
+version = "0.10.0"
 source = { editable = "packages/sie_audio_prep" }
 
 [[package]]
 name = "sie-chroma"
-version = "0.9.0"
+version = "0.10.0"
 source = { editable = "integrations/sie_chroma" }
 dependencies = [
     { name = "chromadb" },
@@ -4504,7 +4504,7 @@ requires-dist = [
 
 [[package]]
 name = "sie-config"
-version = "0.9.0"
+version = "0.10.0"
 source = { editable = "packages/sie_config" }
 dependencies = [
     { name = "fastapi" },
@@ -4563,7 +4563,7 @@ dev = [
 
 [[package]]
 name = "sie-crewai"
-version = "0.9.0"
+version = "0.10.0"
 source = { editable = "integrations/sie_crewai" }
 dependencies = [
     { name = "crewai" },
@@ -4588,7 +4588,7 @@ requires-dist = [
 
 [[package]]
 name = "sie-dspy"
-version = "0.9.0"
+version = "0.10.0"
 source = { editable = "integrations/sie_dspy" }
 dependencies = [
     { name = "dspy" },
@@ -4613,7 +4613,7 @@ requires-dist = [
 
 [[package]]
 name = "sie-haystack"
-version = "0.9.0"
+version = "0.10.0"
 source = { editable = "integrations/sie_haystack" }
 dependencies = [
     { name = "haystack-ai" },
@@ -4640,7 +4640,7 @@ requires-dist = [
 
 [[package]]
 name = "sie-lancedb"
-version = "0.9.0"
+version = "0.10.0"
 source = { editable = "integrations/sie_lancedb" }
 dependencies = [
     { name = "lancedb" },
@@ -4667,7 +4667,7 @@ requires-dist = [
 
 [[package]]
 name = "sie-langchain"
-version = "0.9.0"
+version = "0.10.0"
 source = { editable = "integrations/sie_langchain" }
 dependencies = [
     { name = "langchain-core" },
@@ -4694,7 +4694,7 @@ requires-dist = [
 
 [[package]]
 name = "sie-llamaindex"
-version = "0.9.0"
+version = "0.10.0"
 source = { editable = "integrations/sie_llamaindex" }
 dependencies = [
     { name = "llama-index-core" },
@@ -4721,7 +4721,7 @@ requires-dist = [
 
 [[package]]
 name = "sie-mcp"
-version = "0.9.0"
+version = "0.10.0"
 source = { editable = "packages/sie_mcp" }
 dependencies = [
     { name = "jsonschema" },
@@ -4760,7 +4760,7 @@ dev = [
 
 [[package]]
 name = "sie-qdrant"
-version = "0.9.0"
+version = "0.10.0"
 source = { editable = "integrations/sie_qdrant" }
 dependencies = [
     { name = "qdrant-client" },
@@ -4783,7 +4783,7 @@ requires-dist = [
 
 [[package]]
 name = "sie-sdk"
-version = "0.9.0"
+version = "0.10.0"
 source = { editable = "packages/sie_sdk" }
 dependencies = [
     { name = "aiohttp" },
@@ -4840,7 +4840,7 @@ requires-dist = [
 
 [[package]]
 name = "sie-server"
-version = "0.9.0"
+version = "0.10.0"
 source = { editable = "packages/sie_server" }
 dependencies = [
     { name = "blake3" },
@@ -4981,7 +4981,7 @@ dev = []
 
 [[package]]
 name = "sie-weaviate"
-version = "0.9.0"
+version = "0.10.0"
 source = { editable = "integrations/sie_weaviate" }
 dependencies = [
     { name = "sie-sdk" },