michalharakal · Oct 16, 2026 · Oct 16, 2026 · Oct 16, 2026 · Oct 16, 2026 · Oct 16, 2026
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -11,6 +11,20 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+## [0.61.0] — 2026-10-16
+
+**Padded batches for the encoders.** Against engine **0.60.0**, which gives
+`scaledDotProductAttention` key lengths, a window and a right context carried as op metadata
+(`maskKind`) and a `skainet.sdpa` composite for fused attention kernels. BERT-class encoders and the
+Laya decision models run the sequences of one request as right-padded batches (#490): one attention
+node per layer takes the batch's lengths instead of a dense `[b, 1, S, S]` mask, the lengths are a
+graph input on the compiled path, and the Laya adapter reports `batchSupport = true`. Parity holds
+(the multilingual predict fixtures reproduce through the batched path, 55 values, max |diff| 4.99e-5;
+batched vs single logits within 1.9e-6). On the eager FP32 CPU path a batch saves about 10 % per
+question at best, since that path is compute-bound; the gain is for the compiled lane and for
+accelerators. The BOM re-exports `skainet-bom:0.60.0`. Engine 0.60.0 made its io-core compatibility
+aliases errors; the transformers SKEEP-006 schedule moves by one more minor (see *Deprecated*).
+
 ### Added
 
 - **Padded batches for the encoders** (#490): `PaddedBatch` (`transformer-core`) describes a
@@ -36,6 +50,19 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - **Engine 0.60.0** (`scaledDotProductAttention` with `keyLengths`, `window`, `rightContext`;
   `skainet.sdpa` composite). The version catalog moves `sk.ainet.core:*` to 0.60.0.
 
+### Deprecated
+
+- **Schedule moved by one more minor.** The SKEEP-006 legacy surface (`GGUFTokenizer` and its
+  strategies, `TokenizerType`, `SentencePieceSpecialTokens`, `TekkenTokenizerAdapter`, `ModelRegistry`,
+  the `sk.ainet.apps.kllama.GGUFTokenizer` alias) stays at WARNING in 0.61, goes to ERROR in 0.62
+  and leaves in 0.63; `TokenizerFactory` and the `sk.ainet.apps.llm.Tokenizer` sub-interface are
+  deprecated in 0.62 and removed in 0.63. The runtimes, CLIs and benchmarks still reach the engine
+  through the factory; this release carries the engine bump and the batched encoders, and that
+  migration stays a release of its own. The engine's own io-core aliases (`sk.ainet.io.tokenizer.*`,
+  `sk.ainet.io.model.ModelFormat`, the weight name maps) are errors since engine 0.60.0; the two
+  tests here that still imported `sk.ainet.io.tokenizer.SentencePieceTokenizer` use the
+  `sk.ainet.frontend.tokenizer` home.
+
 ## [0.60.0] — 2026-10-15
 
 **Exact GELU from the engine.** Against engine **0.59.0**, which makes `erf` a first-class op and

diff --git a/README.md b/README.md
@@ -113,7 +113,11 @@ Honest status — see the project-status note at the top of this README.
 
 ## Current release
 
-The current release is **0.60.0**, against **SKaiNET 0.59.0**.
+The current release is **0.61.0**, against **SKaiNET 0.60.0**.
+
+**Padded batches for the encoders.** SKaiNET 0.60.0 gives `scaledDotProductAttention` key lengths, a window and a right context carried as op metadata (`maskKind`), with a `skainet.sdpa` composite for targets that have a fused attention kernel. BERT-class encoders and the Laya decision models run the sequences of one request as right-padded batches (#490): `BertEncoderRuntime.encodeBatch` / `encodePadded`, `LayaDecisionModel.logitsBatch`, the Laya adapter with `batchSupport = true`. One attention node per layer takes the batch's lengths instead of a dense mask, so the compiled path gets the lengths as a graph input. The multilingual predict fixtures reproduce through the batched path (55 values, max |diff| 4.99e-5) and batched and single logits agree within 1.9e-6. On the eager FP32 CPU path a batch saves about 10 % per question at best; the gain is for the compiled lane and accelerators. The SKEEP-006 legacy tokenizer surface stays at WARNING in 0.61; the schedule moves by one more minor (ERROR in 0.62, gone in 0.63).
+
+It builds on **0.60.0**, against **SKaiNET 0.59.0**:
 
 **Exact GELU from the engine.** SKaiNET 0.59.0 makes `erf` a first-class op and gives `gelu` an `approximate` flag that travels as op metadata through the graph executor, the backward rule and the StableHLO export. Laya's decision models now run their exact GELU through `gelu(approximate = false)` instead of a host-side loop per FFN per layer, and the multilingual predict fixtures still reproduce (55 values, max |diff| 4.99e-5). The SKEEP-006 legacy tokenizer surface stays at WARNING in 0.60; the schedule moves by one more minor (ERROR in 0.61, gone in 0.62).
 
@@ -291,7 +295,7 @@ The recommended way to consume is via the BOM. It pins every published `skainet-
 
 ```kotlin
 dependencies {
-    implementation(platform("sk.ainet.transformers:skainet-transformers-bom:0.60.0"))
+    implementation(platform("sk.ainet.transformers:skainet-transformers-bom:0.61.0"))
 
     // Versions resolved from the BOM:
     implementation("sk.ainet.transformers:skainet-transformers-core")

diff --git a/docs/modules/ROOT/pages/how-to/export-moonshine-v2-streaming.adoc b/docs/modules/ROOT/pages/how-to/export-moonshine-v2-streaming.adoc
@@ -32,7 +32,7 @@ From your own build, resolve the published module and run its main class:
 ----
 val exportTool by configurations.creating
 dependencies {
-    exportTool(platform("sk.ainet.transformers:skainet-transformers-bom:0.60.0"))
+    exportTool(platform("sk.ainet.transformers:skainet-transformers-bom:0.61.0"))
     exportTool("sk.ainet.transformers:skainet-transformers-inference-moonshine")
 }
 tasks.register<JavaExec>("exportMoonshine") {

diff --git a/docs/modules/ROOT/pages/reference/moonshine-encoder.adoc b/docs/modules/ROOT/pages/reference/moonshine-encoder.adoc
@@ -24,7 +24,7 @@ a DSL decoder is future work.
 [source,kotlin]
 ----
 dependencies {
-    implementation(platform("sk.ainet.transformers:skainet-transformers-bom:0.60.0"))
+    implementation(platform("sk.ainet.transformers:skainet-transformers-bom:0.61.0"))
     implementation("sk.ainet.transformers:skainet-transformers-inference-moonshine")
 }
 ----

diff --git a/docs/modules/ROOT/pages/reference/tokenizer-api.adoc b/docs/modules/ROOT/pages/reference/tokenizer-api.adoc
@@ -87,7 +87,8 @@ descriptor with `+`. `GGUFModelInfo.identity` is the string identity
 |===
 
 `HuggingFaceBPETokenizer` (no callers since 0.58) was removed in 0.59.0. The schedule for the
-rest moved by one minor in 0.59.0 and by one more in 0.60.0: the legacy classes stay at WARNING
-through 0.60, go to ERROR in 0.61 and leave in 0.62, because the runtimes, CLIs and benchmarks
-still reach the engine through `TokenizerFactory` and the sub-interface, and moving them is a
-release of its own.
+rest moved by one minor in 0.59.0, in 0.60.0 and again in 0.61.0: the legacy classes stay at
+WARNING through 0.61, go to ERROR in 0.62 and leave in 0.63, because the runtimes, CLIs and
+benchmarks still reach the engine through `TokenizerFactory` and the sub-interface, and moving
+them is a release of its own. The engine's io-core aliases they wrapped are errors since
+SKaiNET 0.60.0; new code imports the `sk.ainet.frontend.tokenizer` home directly.
diff --git a/docs/modules/ROOT/pages/reference/typed-decisions.adoc b/docs/modules/ROOT/pages/reference/typed-decisions.adoc
@@ -12,7 +12,7 @@ below put that on the JVM path with the same contract for every model.
 [source,kotlin]
 ----
 dependencies {
-    implementation(platform("sk.ainet.transformers:skainet-transformers-bom:0.60.0"))
+    implementation(platform("sk.ainet.transformers:skainet-transformers-bom:0.61.0"))
     implementation("sk.ainet.transformers:skainet-transformers-decisions-api")      // the contract, no engine dependency
     implementation("sk.ainet.transformers:skainet-transformers-decisions-runtime")  // validation + acceptance policy
     implementation("sk.ainet.transformers:skainet-transformers-inference-laya")     // Laya as a DecisionAdapter

diff --git a/docs/modules/ROOT/pages/tutorials/android-getting-started.adoc b/docs/modules/ROOT/pages/tutorials/android-getting-started.adoc
@@ -55,7 +55,7 @@ dependencies {
     // self-registers it on ART at process start — nothing to call.
     runtimeOnly("sk.ainet.core:skainet-backend-jni-cpu")
 
-    implementation(platform("sk.ainet.transformers:skainet-transformers-bom:0.60.0"))
+    implementation(platform("sk.ainet.transformers:skainet-transformers-bom:0.61.0"))
     implementation("sk.ainet.transformers:skainet-transformers-core")
     implementation("sk.ainet.transformers:skainet-transformers-runtime-kllama")
     implementation("sk.ainet.transformers:skainet-transformers-inference-llama")

diff --git a/docs/modules/ROOT/pages/tutorials/getting-started-java.adoc b/docs/modules/ROOT/pages/tutorials/getting-started-java.adoc
@@ -25,7 +25,7 @@ In your `build.gradle.kts`:
 [source,kotlin]
 ----
 dependencies {
-    implementation(platform("sk.ainet.transformers:skainet-transformers-bom:0.60.0"))
+    implementation(platform("sk.ainet.transformers:skainet-transformers-bom:0.61.0"))
 
     implementation("sk.ainet.transformers:skainet-transformers-runtime-kllama")
     implementation("sk.ainet.transformers:skainet-transformers-agent")
@@ -41,7 +41,7 @@ Or in Maven (Maven needs the `-jvm` classifier suffix on platform artifacts):
     <dependency>
       <groupId>sk.ainet.transformers</groupId>
       <artifactId>skainet-transformers-bom</artifactId>
-      <version>0.60.0</version>
+      <version>0.61.0</version>
       <type>pom</type>
       <scope>import</scope>
     </dependency>

diff --git a/docs/modules/ROOT/pages/tutorials/getting-started-leaf.adoc b/docs/modules/ROOT/pages/tutorials/getting-started-leaf.adoc
@@ -34,7 +34,7 @@ encoder output to the advertised dimensionality. The runtime applies it automati
 [source,kotlin]
 ----
 dependencies {
-    implementation(platform("sk.ainet.transformers:skainet-transformers-bom:0.60.0"))
+    implementation(platform("sk.ainet.transformers:skainet-transformers-bom:0.61.0"))
     implementation("sk.ainet.transformers:skainet-transformers-providers")
 }
 ----

diff --git a/docs/modules/ROOT/pages/tutorials/llama3-tool-calling.adoc b/docs/modules/ROOT/pages/tutorials/llama3-tool-calling.adoc
@@ -52,7 +52,7 @@ The pieces you need live in three modules:
 [source,kotlin]
 ----
 dependencies {
-    implementation(platform("sk.ainet.transformers:skainet-transformers-bom:0.60.0"))
+    implementation(platform("sk.ainet.transformers:skainet-transformers-bom:0.61.0"))
 
     implementation("sk.ainet.transformers:skainet-transformers-runtime-kllama")
     implementation("sk.ainet.transformers:skainet-transformers-agent")

diff --git a/gradle.properties b/gradle.properties
@@ -1,5 +1,5 @@
 GROUP=sk.ainet.transformers
-VERSION_NAME=0.60.0
+VERSION_NAME=0.61.0
 
 POM_DESCRIPTION=SKaiNET-transformers