chore(deps): bump torch

Bumps the pip group with 1 update in the /backend/python/vllm directory: torch.


Updates `torch` from 2.9.1+cpu to 2.12.0+cpu

---
updated-dependencies:
- dependency-name: torch
  dependency-version: 2.12.0+cpu
  dependency-type: direct:production
  dependency-group: pip
...

Signed-off-by: dependabot[bot] <support@github.com>

backend/go/locate-anything-cpp/Makefile

@@ -10,7 +10,7 @@ JOBS?=$(shell nproc --ignore=1)
 # this on `master` always picks up the latest C-API surface (incl. the
 # per-detection accessor functions used by golocateanythingcpp.go).
 LOCATEANYTHING_REPO?=https://github.com/mudler/locate-anything.cpp.git
-LOCATEANYTHING_VERSION?=60e450945476d5e97e0754a8c0e71a9ea81690e0
+LOCATEANYTHING_VERSION?=92c1682da792c1e8a5dec91acc2be4b02c742ded
 
 ifeq ($(NATIVE),false)
 	CMAKE_ARGS+=-DGGML_NATIVE=OFF

backend/index.yaml

@@ -1163,11 +1163,11 @@
 - &opus
   name: "opus"
   alias: "opus"
-  uri: "quay.io/go-skynet/local-ai-backends:latest-cpu-opus"
+  capabilities:
+    default: "cpu-opus"
+    metal: "metal-opus"
   urls:
     - https://opus-codec.org/
-  mirrors:
-    - localai/localai-backends:latest-cpu-opus
   license: BSD-3-Clause
   description: |
     Opus audio codec backend for encoding and decoding audio.
@@ -1177,7 +1177,11 @@
     - opus
     - WebRTC
     - realtime
-    - CPU
+- !!merge <<: *opus
+  name: "opus-development"
+  capabilities:
+    default: "cpu-opus-development"
+    metal: "metal-opus-development"
 - &silero-vad
   name: "silero-vad"
   uri: "quay.io/go-skynet/local-ai-backends:latest-cpu-silero-vad"
@@ -1603,7 +1607,12 @@
   mirrors:
     - localai/localai-backends:master-metal-darwin-arm64-local-store
 - !!merge <<: *opus
-  name: "opus-development"
+  name: "cpu-opus"
+  uri: "quay.io/go-skynet/local-ai-backends:latest-cpu-opus"
+  mirrors:
+    - localai/localai-backends:latest-cpu-opus
+- !!merge <<: *opus
+  name: "cpu-opus-development"
   uri: "quay.io/go-skynet/local-ai-backends:master-cpu-opus"
   mirrors:
     - localai/localai-backends:master-cpu-opus

backend/python/neutts/requirements-cpu.txt

@@ -1,6 +1,7 @@
 --extra-index-url https://download.pytorch.org/whl/cpu
 accelerate
 torch==2.8.0
+torchaudio==2.8.0
 transformers==4.56.1
 librosa==0.11.0
 neucodec>=0.0.4

backend/python/neutts/requirements-cublas12.txt

@@ -3,6 +3,7 @@ neucodec>=0.0.4
 phonemizer==3.3.0
 soundfile==0.13.1
 torch==2.8.0
+torchaudio==2.8.0
 transformers==4.56.1
 resemble-perth==1.0.1
 accelerate

backend/python/vllm/requirements-cpu.txt

@@ -1,6 +1,6 @@
 --extra-index-url https://download.pytorch.org/whl/cpu
 accelerate
-torch==2.9.1+cpu
+torch==2.12.0+cpu
 torchvision
 torchaudio
 transformers

core/backend/options.go

@@ -307,11 +307,19 @@ func gRPCPredictOpts(c config.ModelConfig, modelPath string) *pb.PredictOptions
 		}
 	}
 
+	// TopK may be nil after SetDefaults for backends that don't use llama.cpp's
+	// top_k=40 default (issue #6632, e.g. mlx). proto3 int32 can't be unset, so
+	// send 0 — the value mlx actually wants (top-k disabled).
+	var topK int32
+	if c.TopK != nil {
+		topK = int32(*c.TopK)
+	}
+
 	pbOpts := &pb.PredictOptions{
 		Temperature:         float32(*c.Temperature),
 		TopP:                float32(*c.TopP),
 		NDraft:              c.NDraft,
-		TopK:                int32(*c.TopK),
+		TopK:                topK,
 		MinP:                float32(*c.MinP),
 		Tokens:              int32(*c.Maxtokens),
 		Threads:             int32(*c.Threads),

core/config/backend_capabilities.go

@@ -517,6 +517,33 @@ func NormalizeBackendName(backend string) string {
 	return strings.ReplaceAll(backend, ".", "-")
 }
 
+// nonLlamaSamplerBackends lists backends whose native sampler defaults differ
+// from llama.cpp's, so LocalAI must NOT inject llama.cpp's top_k=40 default for
+// them (issue #6632). mlx_lm's intended default is top_k=0 (disabled) and mlx
+// does not remap 0->40, so shipping 40 silently changes sampling for clients
+// that omit top_k. Leaving TopK nil lets the wire value default to 0.
+//
+// This is intentionally a small allow-list of KNOWN non-llama backends: empty
+// and unknown backends fall through to the llama.cpp default to preserve the
+// GGUF auto-detect path's behavior.
+var nonLlamaSamplerBackends = map[string]struct{}{
+	"mlx":             {},
+	"mlx-vlm":         {},
+	"mlx-distributed": {},
+}
+
+// UsesLlamaSamplerDefaults reports whether a backend should receive llama.cpp's
+// sampler defaults (e.g. top_k=40). Empty/unknown backends return true so the
+// GGUF auto-detect path (which resolves to llama.cpp) keeps today's behavior;
+// only the known non-llama backends in nonLlamaSamplerBackends return false.
+func UsesLlamaSamplerDefaults(backend string) bool {
+	if backend == "" {
+		return true
+	}
+	_, isNonLlama := nonLlamaSamplerBackends[NormalizeBackendName(backend)]
+	return !isNonLlama
+}
+
 // GetBackendCapability returns the capability info for a backend, or nil if unknown.
 // Handles backend name normalization.
 func GetBackendCapability(backend string) *BackendCapability {

core/config/model_config.go

@@ -867,7 +867,12 @@ func (cfg *ModelConfig) SetDefaults(opts ...ConfigLoaderOption) {
 		cfg.Seed = &defaultSeed
 	}
 
-	if cfg.TopK == nil {
+	// top_k=40 is llama.cpp's sampling default and is wrong for backends whose
+	// native default differs (issue #6632). Only inject it for the llama.cpp
+	// family and the empty/auto backend; leave TopK nil for known non-llama
+	// backends (e.g. mlx, whose intended default is top_k=0) so the wire value
+	// is 0 rather than a silently-changed 40.
+	if cfg.TopK == nil && UsesLlamaSamplerDefaults(cfg.Backend) {
 		cfg.TopK = &defaultTopK
 	}
 

core/config/model_config_test.go

@@ -529,4 +529,72 @@ concurrency_groups:
 				"models that template in Go still rely on the Go-generated grammar")
 		})
 	})
+
+	// The default top_k=40 is llama.cpp's sampling default and is WRONG for
+	// backends whose native default differs. mlx_lm's intended default is
+	// top_k=0 (disabled) and mlx does not remap 0->40, so injecting 40 silently
+	// changes sampling for mlx clients that omit top_k (issue #6632). Gate the
+	// injection on backend family: keep 40 for the llama.cpp family and for the
+	// empty/auto backend (the GGUF auto-detect path resolves to llama.cpp), but
+	// leave TopK nil for the mlx family so the wire value is 0.
+	Context("TopK default is backend-gated (issue #6632)", func() {
+		It("injects top_k=40 for the llama.cpp backend", func() {
+			cfg := &ModelConfig{}
+			cfg.Backend = "llama-cpp"
+
+			cfg.SetDefaults()
+
+			Expect(cfg.TopK).NotTo(BeNil(), "llama.cpp must keep its top_k=40 default")
+			Expect(*cfg.TopK).To(Equal(40))
+		})
+
+		It("injects top_k=40 for the empty/auto backend (GGUF auto-detect)", func() {
+			cfg := &ModelConfig{}
+
+			cfg.SetDefaults()
+
+			Expect(cfg.TopK).NotTo(BeNil(), "empty backend resolves to llama.cpp; default unchanged")
+			Expect(*cfg.TopK).To(Equal(40))
+		})
+
+		It("leaves TopK nil for the mlx backend", func() {
+			cfg := &ModelConfig{}
+			cfg.Backend = "mlx"
+
+			cfg.SetDefaults()
+
+			Expect(cfg.TopK).To(BeNil(),
+				"mlx_lm's intended default is top_k=0 (disabled); LocalAI must not inject 40")
+		})
+
+		It("leaves TopK nil for the mlx-vlm backend", func() {
+			cfg := &ModelConfig{}
+			cfg.Backend = "mlx-vlm"
+
+			cfg.SetDefaults()
+
+			Expect(cfg.TopK).To(BeNil())
+		})
+
+		It("leaves TopK nil for the mlx-distributed backend", func() {
+			cfg := &ModelConfig{}
+			cfg.Backend = "mlx-distributed"
+
+			cfg.SetDefaults()
+
+			Expect(cfg.TopK).To(BeNil())
+		})
+
+		It("respects an explicit top_k even for the mlx backend", func() {
+			explicit := 7
+			cfg := &ModelConfig{}
+			cfg.Backend = "mlx"
+			cfg.TopK = &explicit
+
+			cfg.SetDefaults()
+
+			Expect(cfg.TopK).NotTo(BeNil())
+			Expect(*cfg.TopK).To(Equal(7))
+		})
+	})
 })

core/http/endpoints/openai/realtime.go

@@ -990,8 +990,18 @@ func updateSession(session *Session, update *types.SessionUnion, cl *config.Mode
 	}
 
 	if rt.Audio != nil && rt.Audio.Input != nil && rt.Audio.Input.Transcription != nil {
-		session.InputAudioTranscription = rt.Audio.Input.Transcription
-		session.ModelConfig.Pipeline.Transcription = rt.Audio.Input.Transcription.Model
+		trUpd := rt.Audio.Input.Transcription
+		// A language-only update (e.g. a client forcing the STT language) carries
+		// an empty Model. Preserve the pipeline's configured transcription backend
+		// instead of blanking it — otherwise the next utterance transcribes against
+		// an empty model and the backend RPC fails with "unimplemented".
+		if trUpd.Model == "" && session.InputAudioTranscription != nil {
+			trUpd.Model = session.InputAudioTranscription.Model
+		}
+		session.InputAudioTranscription = trUpd
+		if trUpd.Model != "" {
+			session.ModelConfig.Pipeline.Transcription = trUpd.Model
+		}
 	}
 
 	if rt.Model != "" || (rt.Audio != nil && rt.Audio.Output != nil && rt.Audio.Output.Voice != "") || (rt.Audio != nil && rt.Audio.Input != nil && rt.Audio.Input.Transcription != nil) {

go.mod

@@ -36,7 +36,7 @@ require (
 	github.com/mholt/archiver/v3 v3.5.1
 	github.com/microcosm-cc/bluemonday v1.0.27
 	github.com/modelcontextprotocol/go-sdk v1.5.0
-	github.com/mudler/cogito v0.9.5-0.20260315222927-63abdec7189b
+	github.com/mudler/cogito v0.10.1-0.20260609212329-bf4010d31047
 	github.com/mudler/edgevpn v0.34.0
 	github.com/mudler/go-processmanager v0.1.1
 	github.com/mudler/memory v0.0.0-20260406210934-424c1ecf2cf8

go.sum

@@ -968,8 +968,8 @@ github.com/mschoch/smat v0.2.0 h1:8imxQsjDm8yFEAVBe7azKmKSgzSkZXDuKkSq9374khM=
 github.com/mschoch/smat v0.2.0/go.mod h1:kc9mz7DoBKqDyiRL7VZN8KvXQMWeTaVnttLRXOlotKw=
 github.com/mudler/LocalAGI v0.0.0-20260606071251-14aed1ae4336 h1:iKBkSnpisOvMVxFoYsAObvAuOqXBakRPMD0PWxWG5EE=
 github.com/mudler/LocalAGI v0.0.0-20260606071251-14aed1ae4336/go.mod h1:U+g6u8mF2wQxhkdBl3dr8G4db1cv3n7KTKmraoJ7D0c=
-github.com/mudler/cogito v0.9.5-0.20260315222927-63abdec7189b h1:A74T2Lauvg61KodYqsjTYDY05kPLcW+efVZjd23dghU=
-github.com/mudler/cogito v0.9.5-0.20260315222927-63abdec7189b/go.mod h1:6sfja3lcu2nWRzEc0wwqGNu/eCG3EWgij+8s7xyUeQ4=
+github.com/mudler/cogito v0.10.1-0.20260609212329-bf4010d31047 h1:wJ8WbDah1YcpBNRDmovQro8JiR228YFk7TUqPCS4m04=
+github.com/mudler/cogito v0.10.1-0.20260609212329-bf4010d31047/go.mod h1:6sfja3lcu2nWRzEc0wwqGNu/eCG3EWgij+8s7xyUeQ4=
 github.com/mudler/edgevpn v0.34.0 h1:qDrD/rCPFY/FdURbXudIZWihVKY4VOX3nMn3CcbeQEU=
 github.com/mudler/edgevpn v0.34.0/go.mod h1:yki7uMi5LR9gSMrw8PdPieuxsrk8BLV2Ui7VBEmbbIA=
 github.com/mudler/go-piper v0.0.0-20241023091659-2494246fd9fc h1:RxwneJl1VgvikiX28EkpdAyL4yQVnJMrbquKospjHyA=