* fix(router): score classifier production-readiness
Conversation trimming runs through the classifier model's chat template
and trims by exact token count, sized to the model's n_batch which is
now scaled to context so long probes can't crash the backend. Missing
chat_message templates are a hard error at router build time. Router-
facing factories (Embedder/Scorer/Reranker/TokenCounter) re-resolve
ModelConfig per call so a model installed post-startup doesn't bind a
stub Backend="" config and silently fall into the loader's auto-
iterate path.
New 'vector_store' backend trace recorded inside localVectorStore on
every Search/Insert — including the backend-load-failure path that
previously vanished into an xlog.Warn — with outcome tagging
(hit/miss/empty_store/backend_load_error/find_error/insert_error/ok).
Companion cleanup drops misleading similarity:0 and input_tokens_count:0
from non-hit and text-mode traces.
Gallery local-store-development aliases to 'local-store' so the master
image satisfies pkg/model.LocalStoreBackend lookups from the embedding
cache.
Misc: llama-cpp TokenizeString reads the correct 'prompt' JSON key
(the original bug); ModelTokenize nil-guard; non-fatal mitm proxy
startup; PII 'route_local' renamed to 'allow' with docs/UI in sync;
model-editor footer no longer eats the edit area on small screens;
several config-editor template/dropdown/section fixes.
Tests: e2e router specs (casual/code-hint + long-conversation trim),
vector_store trace specs, lazy-factory specs, gallery dev-alias
resolution, Playwright trace badge + scroll regression.
Assisted-by: Claude:claude-opus-4-7 [Claude Code]
Signed-off-by: Richard Palethorpe <io@richiejp.com>
* feat(backend): auto-size batch to context for embedding and rerank models
Embedding and rerank models pool over the whole input in a single physical batch (n_ubatch). With batch left at the 512 default, the backend rejects longer inputs with "input is too large to process", silently capping a large-context embedder (e.g. 8k/32k) at 512 tokens. Size n_batch to the context for these single-pass usecases, mirroring the existing FLAG_SCORE behaviour; an explicit batch: still wins.
Extracts EffectiveContextSize/EffectiveBatchSize from grpcModelOpts so the effective decode window has one home for other callers to reuse.
Adds an e2e-aio regression test that embeds a >512-token input. The AIO embedding model is switched to nomic-embed-text-v1.5 (2048 context) because the previous granite model was capped at 512 tokens and could not exercise the larger batch.
Assisted-by: claude-code:claude-opus-4-8 [Claude Code]
Signed-off-by: Richard Palethorpe <io@richiejp.com>
* fix(gallery): raise arch-router scoring output cap via parallel:64
Scoring decodes the whole prompt+candidate in a single llama_decode and
reads one logit row per candidate token. The vendored llama.cpp server
caps causal output rows at n_parallel, so the default of 1 aborts with
GGML_ASSERT(n_outputs_max <= cparams.n_outputs_max) on multi-token route
labels. Set options: [parallel:64] on both arch-router quant entries to
lift the cap; kv_unified (the grpc-server default) keeps the full context
per sequence, so this does not split the KV cache.
Assisted-by: claude-code:claude-opus-4-8 [Claude Code]
Signed-off-by: Richard Palethorpe <io@richiejp.com>
---------
Signed-off-by: Richard Palethorpe <io@richiejp.com>
Add a routing middleware stack and a cloud-proxy backend.
* cloud-proxy: a Go gRPC backend that forwards OpenAI- and
Anthropic-shaped chat requests to upstream providers, with an
optional translate mode (OpenAI request -> Anthropic /v1/messages
-> OpenAI response) and full tool-calling support.
* routing: admission control, content-aware model routing
(embedding cache + classifier + rerank + Arch-Router score),
PII detection/redaction (regex + NER) with streaming filter and
OpenAI/Anthropic adapters, and a per-user/per-key billing recorder
backed by GORM or in-memory storage.
* middleware: UsageMiddleware records usage via the billing recorder,
plus admission, route-model, usage-stamp and trace middlewares.
* observability: BackendTrace ring buffer stores full request bodies
(capped), MITM proxy emits structured trace events, and router
classifier decisions surface at /api/router/decide.
* gallery: Arch-Router-1.5B (Q4_K_M and Q8_0).
* UI: cloud-proxy model-editor fields, classifier system-prompt and
score-normalization config, and a Traces page rendering request
bodies.
Assisted-by: claude-code:claude-opus-4-7 [Read] [Edit] [Bash]
Signed-off-by: Richard Palethorpe <io@richiejp.com>
* fix(streaming): comply with OpenAI usage / stream_options spec (#8546)
LocalAI emitted `"usage":{"prompt_tokens":0,...}` on every streamed
chunk because `OpenAIResponse.Usage` was a value type without
`omitempty`. The official OpenAI Node SDK and its consumers
(continuedev/continue, Kilo Code, Roo Code, Zed, IntelliJ Continue)
filter on a truthy `result.usage` to detect the trailing usage chunk;
LocalAI's zero-but-non-null usage on every intermediate chunk made
that filter swallow every content chunk and surface an empty chat
response while the server log looked successful.
Changes:
- `core/schema/openai.go`: `Usage *OpenAIUsage \`json:"usage,omitempty"\``
so intermediate chunks no longer carry a `usage` key. Add
`OpenAIRequest.StreamOptions` with `include_usage` to mirror OpenAI's
request field.
- `core/http/endpoints/openai/chat.go` and `completion.go`: keep using
the `Usage` struct field as an in-process channel for the running
cumulative, but strip it before JSON marshalling. When the request
set `stream_options.include_usage: true`, emit a dedicated trailing
chunk with `"choices": []` and the populated usage (matching the
OpenAI spec and llama.cpp's server behavior).
- `chat_emit.go`: new `streamUsageTrailerJSON` helper; drop the
`usage` parameter from `buildNoActionFinalChunks` since chunks no
longer carry usage.
- Update `image.go`, `inpainting.go`, `edit.go` to wrap their Usage
values with `&` for the new pointer field.
- UI: send `stream_options:{include_usage:true}` from the React
(`useChat.js`) and legacy (`static/chat.js`) chat clients so the
token-count badge keeps populating now that the server is
spec-compliant.
Tests:
- New `chat_stream_usage_test.go` pins the spec invariants:
intermediate chunks have no `usage` key, the trailer JSON has
`"choices":[]` and a populated `usage`, and `OpenAIRequest` parses
`stream_options.include_usage`.
- Update `chat_emit_test.go` to reflect that finals no longer embed
usage.
Verified against the live LocalAI instance: before the fix Continue's
filter logic swallowed 16/16 token chunks; with the new shape it
yields 4/5 and routes usage through the dedicated trailer chunk.
Fixes#8546
Assisted-by: Claude:opus-4.7 [Claude Code]
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* fix(streaming): silence errcheck on usage trailer Fprintf
The new spec-compliant `stream_options.include_usage` trailer writes
were flagged by errcheck since they're new code (golangci-lint runs
new-from-merge-base on master); the surrounding `fmt.Fprintf` data:
writes are grandfathered. Drop the return values explicitly to match
the linter's contract without adding a nolint shim.
Assisted-by: Claude:opus-4.7 [Claude Code]
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
---------
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
Co-authored-by: Ettore Di Giacinto <mudler@localai.io>
* feat(functions): add peg-based parsing
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* feat: support returning toolcalls directly from backends
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* chore: do run PEG only if backend didn't send deltas
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
---------
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* Initial plan
* Fix SSE streaming format to comply with specification
- Replace json.Encoder with json.Marshal for explicit formatting
- Use explicit \n\n for all SSE messages (instead of relying on implicit newlines)
- Change %v to %s format specifier for proper string formatting
- Fix error message streaming to include proper SSE format
- Ensure consistency between chat.go and completion.go endpoints
Co-authored-by: mudler <2420543+mudler@users.noreply.github.com>
* Add proper error handling for JSON marshal failures in streaming
- Handle json.Marshal errors explicitly in error response paths
- Add fallback simple error message if marshal fails
- Prevents sending 'data: <nil>' on marshal failures
- Addresses code review feedback
Co-authored-by: mudler <2420543+mudler@users.noreply.github.com>
* Fix SSE streaming format to comply with specification
Co-authored-by: mudler <2420543+mudler@users.noreply.github.com>
* Fix finish_reason field to use pointer for proper null handling
- Change FinishReason from string to *string in Choice schema
- Streaming chunks now omit finish_reason (null) instead of empty string
- Final chunks properly set finish_reason to "stop", "tool_calls", etc.
- Remove empty content from initial streaming chunks (only send role)
- Final streaming chunk sends empty delta with finish_reason
- Addresses OpenAI API compliance issues causing client failures
Co-authored-by: mudler <2420543+mudler@users.noreply.github.com>
* Improve code consistency for string pointer creation
- Use consistent pattern: declare variable then take address
- Remove inline anonymous function for better readability
- Addresses code review feedback
Co-authored-by: mudler <2420543+mudler@users.noreply.github.com>
* Move common finish reasons to constants
- Create constants.go with FinishReasonStop, FinishReasonToolCalls, FinishReasonFunctionCall
- Replace all string literals with constants in chat.go, completion.go, realtime.go
- Improves code maintainability and prevents typos
Co-authored-by: mudler <2420543+mudler@users.noreply.github.com>
* Make it build
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* Fix finish_reason to always be present with null or string value
- Remove omitempty from FinishReason field in Choice struct
- Explicitly set FinishReason to nil for all streaming chunks
- Ensures finish_reason appears as null in JSON for streaming chunks
- Final chunks still properly set finish_reason to "stop", "tool_calls", etc.
- Complies with OpenAI API specification example
Co-authored-by: mudler <2420543+mudler@users.noreply.github.com>
---------
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>
Co-authored-by: mudler <2420543+mudler@users.noreply.github.com>
Co-authored-by: Ettore Di Giacinto <mudler@users.noreply.github.com>
Co-authored-by: Ettore Di Giacinto <mudler@localai.io>
* WIP - add endpoint
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* Rename
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* Wire the Completion API
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* Try to make it functional
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* Almost functional
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* Bump golang versions used in tests
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* Add description of the tool
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* Make it working
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* Small optimizations
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* Cleanup/refactor
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* Update docs
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
---------
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
- Add a system backend path
- Refactor and consolidate system information in system state
- Use system state in all the components to figure out the system paths
to used whenever needed
- Refactor BackendConfig -> ModelConfig. This was otherway misleading as
now we do have a backend configuration which is not the model config.
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* migrate core/system to pkg/system - it has no dependencies FROM core, and IS USED in pkg
Signed-off-by: Dave Lee <dave@gray101.com>
* move pkg/templates up to core/templates -- nothing in pkg references it, but it does reference core.
Signed-off-by: Dave Lee <dave@gray101.com>
* remove extra check, len of nil is 0
Signed-off-by: Dave Lee <dave@gray101.com>
* move pkg/startup to core/startup -- it does have important and unfixable dependencies on core
Signed-off-by: Dave Lee <dave@gray101.com>
---------
Signed-off-by: Dave Lee <dave@gray101.com>
Rename LocalAI-Extra-Usage -> Extra-Usage, add MACHINE_TAG as cli flag option, add docs about extra-usage and machine-tag
Signed-off-by: mintyleaf <mintyleafdev@gmail.com>
* Add machine tag option, add extraUsage option, grpc-server -> proto -> endpoint extraUsage data is broken for now
Signed-off-by: mintyleaf <mintyleafdev@gmail.com>
* remove redurant timing fields, fix not working timings output
Signed-off-by: mintyleaf <mintyleafdev@gmail.com>
* use middleware for Machine-Tag only if tag is specified
Signed-off-by: mintyleaf <mintyleafdev@gmail.com>
---------
Signed-off-by: mintyleaf <mintyleafdev@gmail.com>
* Read jinja templates as fallback
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* Move templating out of model loader
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* Test TemplateMessages
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* Set role and content from transformers
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* Tests: be more flexible
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* More jinja
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* Small refactoring and adaptations
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
---------
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
fix(model-list): be consistent, skip known files from listing
This changeset does two things:
- Removes the dependency of listing models from the OpenAI schema.
- Tries to reduce confusion between ListModels() in model loader and in
the service - now there is only one ListModels which is in services
and does not depend anymore on the OpenAI schema
- The OpenAI-schema functions were moved nearby the OpenAI specific
endpoints that needs the schema
- Drops the ListModel Service structure as there was no real need for
it.
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* fix(grammar): Fix JSON mode and custom grammar
* tests(aio): add jsonmode test
* tests(aio): add functioncall test
* fix(aio): use hermes-2-pro-mistral as llm for CPU profile
* add phi-2-orange
* core 1
* api/openai/files fix
* core 2 - core/config
* move over core api.go and tests to the start of core/http
* move over localai specific endpoints to core/http, begin the service/endpoint split there
* refactor big chunk on the plane
* refactor chunk 2 on plane, next step: port and modify changes to request.go
* easy fixes for request.go, major changes not done yet
* lintfix
* json tag lintfix?
* gitignore and .keep files
* strange fix attempt: rename the config dir?
