Repository navigation
0.8.0 — thirteen catalog models; hybrid chats resume from a checkpoint; four models stop at the end of their KV cache
LatestThirteen catalog models join (67 → 80): six decision models on five new kinds, zero-shot text classification, LFM3 1.7B with tool calling, MiniCPM-V 4.7 1B, Fun-ASR-Nano, Audio8-TTS, 8-speaker diarization and quantile forecasting. Chats on a hybrid model resume from a checkpoint instead of prefilling the whole conversation again. VibeVoice, VoxCPM, VoxCPM2 and Unlimited-OCR stop at the end of their KV cache, and a graph compiled for one iPhone is picked only on that iPhone. This is a minor release: some code written for 0.7.3 needs an edit, listed under Changed. A package that depends on from: "0.7.3" gets 0.8.0 the next time its packages are updated (swift package update, or Update to Latest Package Versions in Xcode); one that depends on exact: "0.7.3" stays on 0.7.3 until that line changes. Built and gated on macOS 27.0 (26A428) and Xcode 27 (27A266a).
Start here: installation and smallest working example · Examples/TextClassify · Examples/Week
The systemone command-line tool, from Homebrew:
brew upgrade systemone # from 0.7.3The Swift package, in Package.swift:
.package(url: "https://github.com/john-rocky/coreai-kit", exact: "0.8.0")Changed
- What code written for 0.7.3 may need.
VibeVoicePaths.inBundleDir(_:glueDir:voicesDir:embedTokens:)now throws when the folder holds
only graphs compiled for other devices (under Fixed): call it withtry.- A
switchwithoutdefaultoverCatalogEntry.Kind,Decision.Format,VisionError,
KitBundleError,VibeVoiceErrororVoxCPMError: each has new cases (the kinds and
jointHeadunder Added, the errors under Fixed). Add adefault. - A reference to one of the five functions below by its 0.7.3 full name, without a call (for
examplelet make = SpeakerSegment.init(speaker:startFrame:endFrame:)), no longer compiles: each
gained parameters with default values, so its full name changed. Wrap it in a closure instead
(let make = { SpeakerSegment(speaker: $0, startFrame: $1, endFrame: $2) }). A direct call that
leaves the new parameters out builds as before.SpeakerSegment.init(speaker:startFrame:endFrame:)gainedframeSec.KitDiarizer.segments(from:bridgeFrames:)gainedframeSec.CatalogEntry.init(…)gainedassets.CatalogEntry.Variant.init(path:sizeMB:)gainedengineandarchitectures.VLArchitecture.init(…)gainedropeShiftFrameTailandassistantPrefix.
Added
-
Chats on a hybrid model resume from a checkpoint — on Qwen3.5, LFM2.5, LFM3 and Granite 4 the
recurrent state cannot be cut back, so everyChatSessionturn and everyKitLanguageModelcall used to
re-prefill the whole conversation. Both now ask the engine to checkpoint the state at the end of the prompt's
last finished message (InferenceOptions.checkpointAfter, just past the last turn-end token such as
<|im_end|>; the engine saves it while it prefills, without a call of its own, on the chunk grid of a
chunkedprefill). The next turn rewinds to the checkpoint, or keeps everything when it extends the previous
prompt, and prefills only what follows. On an M4 Max (2026-10-30), turn 2 of a conversation about a
1,584-token document on LFM3-1.7B: 837.8 ms to the first token with the history re-prefilled, 38.6 ms from
the checkpoint; replies identical in both ways on LFM3-1.7B, Qwen3.5-0.8B and LFM2.5-1.2B. On an iPhone 17 Pro,
Qwen3.5-0.8B: 19.8 → 3.2 s on a 484-token conversation, replies identical. A first turn costs nothing
measurable. A template that rewrites the earlier history, and the sequential engine, re-prefill as before. -
coreai-models pinned to
0.2.9-zoo(was0.2.8-zoo) for the checkpoint above. It also brings the padded
last chunk: a bundle whose metadata declares that itsprefillfunction takes a padded chunk
("prefill_pad_tail": true) feeds what is left of a prompt after its full chunks as one call instead of up
to 63 single steps (LFM3-1.7B, 1,000-token prompt: 0.692 → 0.500 s to the first token on an M4 Max). All
lockfiles follow. -
Tool calling for LFM3 —
KitLanguageModeldeclares.toolCallingfor a bundle whose vocab carries LFM3's
<function=/<parameter=tokens and renders the tool turns the way LFM3's chat template does (<tools>lines
injson.dumpsform, XML<tool_call>calls,<tool_response>results, reasoning replayed inside the tool loop),
then parses the model's XML calls with values typed by the tool's schema. The render equals transformers'
apply_chat_templatestring and ids on 30 conversations; LFM2.5 bundles still take no tools. -
Streams that start inside a thinking block — when a rendered prompt ends with an open thinking marker
(LFM3's template with thinking on ends<think>),ChatSessionandKitLanguageModelparse the reply as
thinking until the close marker, so the reasoning no longer reaches the answer text. A prompt that ends anywhere
else parses as before. -
LFM3 1.7B — catalog id
lfm3-1.7b(kind: chat, macOS), the zoo's int8 bundle on the pipelined engine: it
thinks by default and calls tools throughLFM3ToolDialect. -
An engine hint per platform — a catalog variant may name its own
engine(CatalogEntry.Variant.engine), which
wins over the entry's on that platform (CatalogEntry.resolvedEngine);ChatSession(catalog:),TypedDecisionsand
KitTextNormalizerload with it.lfm3-1.7bgains an iPhone variant on the static-shape engine: the Neural Engine
form, compiled per iPhone generation into the repo'sios-<arch>/(ios-h19p/for the iPhone 18 Pro at this upload;
noios/). A kit without the key reads the entry'sengine("pipelined", the Mac bundle's). -
Which iPhones have a bundle, before the download — an
iosvariant in a repo with noios/lists the
architectures it has anios-<arch>/for (CatalogEntry.Variant.architectures;lfm3-1.7blistsh19p).
CatalogEntry.unavailableReasonreads it, and the architecture of a path inside anios-<arch>/folder, against
the device's (AIModel.deviceArchitectureName): nil when the phone has a bundle, the Mac, or a phone Core AI does
not name, else a sentence for a picker.ModelCatalog.entry(forID:), which everycatalog:initializer rides,
throwsunsupportedDevicewith it before any download. A kit without the key ignores it. -
ChatSession.Configuration.enableThinking— rendersenable_thinkinginto the bundle's chat template
(nil, the default, leaves the template's own default);falsegives LFM3 and the Qwen3 family their thinking-off
prompt, so the response budget goes to the answer. On a template that does not read the variable (LFM2.5 2.6B,
Nanbeige4.1 3B),falsecloses an empty think block after the generation prompt instead. -
MiniCPM-V 4.7 1B — catalog id
minicpm-v-4.7-1b(kind: vlm), loaded byKitVisionModelthrough the new
VLModelID.miniCPMV47/VLArchitecture.miniCPMV47: OpenBMB's 4.6 vision tower (448 px, 64 image tokens) in front
of a decoder that places the image on a canvas (M-RoPE over T, H and W) with an untied head, answering with
thinking off. The prompt is the reference processor's —<image_id>0</image_id><image>, the 64 pads,</image>—
and two newVLArchitecturefields, both defaulting to the old behaviour, carry the rest:ropeShiftFrameTail
(</image>and the newline after it sit on the canvas, so the rope shift starts after them) andassistantPrefix
(the empty think block the chat template writes at the start of every assistant turn). The int8hu decoder and the
int8lin tower are the pair the zoo gated on an iPhone 18 Pro (iOS 27.2): 98.0 tok/s decode and 68.2 ms per image
in its bench. -
Timestral-209M — catalog id
timestral-209m(the newkind: quantileForecasting), driven by the new
KitTimestralForecasterand the opCoreAI.forecastQuantiles(_:horizon:options:): a number series in,horizon
steps of 21 quantiles (0.01 to 0.99) out, as aQuantileForecastwithmedianandquantile(_:). By default
each graph runs where its form computes right: the form that is fp16 throughout (fp16p7-qkv4-tanh-conv) on the
Neural Engine, the other forms on the GPU. Asking for any other unit for the fp16-throughout form throws
KitTimestralForecasterError.wrongComputeUnits, and an output at fp16's limit throws.saturatedOutput, instead
of returning a forecast. Download: 1,214 MB on macOS and on iOS; licenseLicenseRef-Timestral-Modified-MIT. The
kind is new: a kit built before it decodes the entry asunknownand leaves it out ofavailable(_:).Examples/Forecastruns TimesFM
2.5 or Timestral-209M on pasted numbers or a CSV file for a horizon you choose, and its CLI takes the same inputs. -
Examples/TextClassify — your own messages, sorted into categories you name. Paste them (one a line, or blocks
separated by a blank line), import a .txt, .md or .csv file (UTF-8; anything else is refused with the reason), or
take the sample inbox; at most 2,000 messages. Name 2 to 12 categories in place of the eight support intents.
Every message still gets an urgency and a sentiment from the same forward, and By category lists each category
with its most urgent messages first. On a Mac, the download, the load and a sort hold a user-initiated activity,
so App Nap no longer slows a download in a background window to 0.1–0.2 MB/s. Measured 2026-10-10: 100 pasted
messages at 30.0 to 31.0 ms median each on an M4 Max and 38.4 to 38.6 ms on an iPhone 18 Pro, every message with
the same three answers as the sample inbox. -
Examples/Week — your own week planned on one tap: every event is one
decider-0.8bdecision (what it
needs before it, seven options), and the events that need something are gathered under Before your week, in
time order. The week comes from your calendar (this week, read only: nothing is ever written to a calendar),
from pasted lines (Mon 11:00 Title · Place · Notes: …, or aweek-cli dump --outJSON file; unreadable
lines are counted and skipped, at most 200 events), or from a seeded sample week of fictional events. Add
reminders asks for Reminders access only when pressed.week-cli rungives the same loop and JSON on a Mac
and gives the app's 20 answers back, event by event. Measured on an M4 Max (macOS 27.0, 2026-10-09): 20
events in 12.90 to 13.01 s, 627 to 651 ms median per event. -
d1-3B — catalog id
d1-3b(the newkind: tokenDecision,format: optionRows), driven by the newKitD1Decider
(Sources/CoreAIKit/D1/): Liquid AI's decision model from LFM2.5-VL-3B (LFM Open License v1.0; a SigLIP2 vision tower
and the LFM2 hybrid decoder). A state (text or JSON), pictures and typed questions (noul, choice, score) go in; every
option of every question comes back with a probability; nothing is generated. Each question is its own row in the
provider's chat form; the decoder (no vocabulary head, a static 64 tokens a call from zeroed states) returns the hidden
state at every position, and the host reads the row's last one against the tied embedding rows of each option's tokens
(head/option_rows, 2,134 ids) in float64, each option's highest logit, a softmax over the options. A picture is
capped at one megapixel, cut into the processor's crops and read by the vision tower once per crop. The host is the
model zoo'sapps/D1library (e36ad15) with every numeric path unchanged: across its eleven files only names
(prefixedD13B; the zoo'sD1DeciderisD13BPipeline), access,@available(macOS 27, iOS 27, *), the Intel-Mac
guard on the three files that hold Float16 and a provenance line differ, and undoing those gives the zoo's files back
byte for byte. Its readout's products go through the BLAS calls NumPy makes, in a new C target,CoreAIKitD1BLAS
(defines only, no unsafe flags; the zoo'sD1BLASwith its function namedcoreaikit_d13b_matvec). Two decoders,
one contract: 5,598 MB for the Mac (fp16) and 3,740 MB for the iPhone (MLP linears int8 per block of 32); the vision
tower (854 MB), named in a newassets.towerfield that a kit built before it ignores, downloads when a request
carries a picture. An iPhone app needscom.apple.developer.kernel.increased-memory-limit(without it the
decoder does not load; the kit cannot add it). A row holds at most 4,032 tokens and a request's pictures at most 2,816
image tokens, refused before any graph call in the provider's words. The kind is new: kits 0.1.0–0.7.3 decode an
unknown kind asunknownand leave the entry out ofavailable(_:); the bundle's metadata sayskind: decision-backbonewith nodecision.head, whichTypedDecisionscannot read, andTypedDecisionsrefuses it by its
kind before downloading anything (soCoreAI.decideandCoreAI.systemOnedo too).systemone serve,systemone mcpandSystemOne.backend(catalog:)loadKitD1Deciderfor the kind (configuration.sharePrefixreaches it), and so
dodecide-cli ask(--image) andserve(--bundle <decoder> [--tower <tower>]).prepare(state:)and
decide(prepared:questionsJSON:)answer later questions on a kept state;readout(requestJSON:images:shared:)gives
each row's ids, hidden digest, logits and p and the provider's response.D13BTests(the catalog entry and the
assets.towerfield, the refusals before a download, the request in the provider's form, the option order, 24 rows'
text and readout against the Python reference without files; with a tokenizer, the zoo's 393 fixture rows against
host.py; with the bundles, 28 questions — 16 records of text and 2 pictures — bit for bit against the Python
reference's run of the same bundle, the shared prefix, a prepared state, the wire and the five refusals, measured on an
M4 Max: 28 of 28 rows and 18 of 18 responses). On the wire, instructions, levels and descriptions given as JSON values
read as their JSON text, the kit's rule for every model (the provider's code takes text only;
readout(requestJSON:)refuses them as it does).systemone serve --model d1-3bpassesconformance/check.py(22
requests and the three routes). -
d1-omni-600M — catalog id
d1-omni-600m(the newkind: omniDecision,format: markerScores), driven by the new
KitD1OmniDecider(Sources/CoreAIKit/D1Omni/): Liquid AI's decision model (LFM Open License v1.0; an
LFM2.5-Encoder-350M trunk with a typed decision head, a SigLIP2 vision tower and a FastConformer audio tower). A state
(text or JSON) with images or one 16 kHz clip, and typed questions (noul, choice, score) go in; every option of every
question comes back with a probability; nothing is generated. Each question is its own row in the publisher's form, a
<|mask|>marker before each option, read from one forward pass of a static fp16 decision graph at the smallest of
seven lengths (64 to 4,096 positions) that holds the row, behind the prefix rows of the vision graph (one call per
image crop) or the audio graph (5 / 10 / 20 / 30 s buckets). The host is the model zoo'sapps/D1Omnilibrary
(f9e0e09) with every numeric path unchanged: across its twelve files only names (prefixedD1; the zoo'sD1Omniis
D1OmniPipeline), access,@available(macOS 27, iOS 27, *)and a provenance line differ, and a request's state is
tokenized once for all its questions. 6,458 MB, the same.aimodels on macOS and iOS: the platform folder's twelve
bundles download whole on first use, each its own download, so an iPhone takes the JITios/graphs and never the
repo'sios-h19p/AOT set (which has no L4096); a graph loads the first time a row or a medium needs it. The kind is
new: kits 0.1.0–0.7.3 decode an unknown kind asunknownand leave the entry out ofavailable(_:)(each tag's
ModelCatalog.swift, read); the bundle's metadata saysdecision.head = "encoder", sodecisionwould put it in
front of laya's encoder path, which cannot read its rows.TypedDecisionsrefuses it by its kind before downloading
anything./v1/systemonegainsaudio(one clip: a data URL, base64, or a file path, whichSystemOneServerreads
only on 127.0.0.1); a request withaudiomay leavestateout, and every backend butKitD1OmniDeciderrefuses
it with a 422 that names the model.systemone serve,systemone mcp(thedecidetool's schema gainsaudio) and
SystemOne.backend(catalog:)loadKitD1OmniDeciderfor the kind, and so dodecide-cli ask(--image,--audio)
andserve(--bundletakes the repo'smacos/folder).D1OmniTests(the catalog entry, the refusals, the wire's
audio, the request in the publisher's form, the readout of 38 fixture rows against the publisher's response text;
with the repo'stokenizer/andreference/, the zoo's 360 public rows token for token; with the bundles, 14
requests — 20 text rows at 64 to 4,096 positions, 3 images, 3 clips — bit for bit against the zoo's Swift host,
measured on an M4 Max: 38 of 38 rows and 14 of 14 responses). -
Kev-0.8B and Kev-4B — catalog ids
kev-0.8bandkev-4b(the newkind: rowDecision,format: pointerHead),
driven by the newKitKevDecider(Sources/CoreAIKit/Kev/): Jared Palmer's decision models, a rank-16 LoRA adapter
and a pointer head on Qwen3.5-0.8B-Base and Qwen3.5-4B-Base (Apache-2.0). A state (text or JSON) and typed questions
(noul, choice, score) go in, and every option of every question comes back with a probability; nothing is generated.
Each question is its own row: the state, then the question and its options between the author's delimiter tokens. The
host is the model zoo'sapps/Kevlibrary (9e06b5a) with every numeric path unchanged: the author's rendering and row
form, the decoder (fp16, no vocabulary head, a static 128 tokens per call, each Gated DeltaNet recurrence in an fp32
Metal kernel) returning the hidden state at every position, and the author's pointer head on the host in float64 at
the author's temperature, each p rounded to float32 once. The shared prefix is on by default: a request's state runs
its whole 128-token calls once, and on this graph every row's hidden state equals the direct run's bit for bit.
prepare(state:)anddecide(prepared:questionsJSON:)answer later questions on a kept state. A choice lists up to 255
options; a row holds at most 3,968 tokens. 1,520 MB for Kev-0.8B, the same.aimodelon macOS and iOS (the zoo gated
it on an iPhone 18 Pro; the kit has not run on a phone), and 8,431 MB for Kev-4B, macOS only. The first load specializes
the graph: 3.3 s for 0.8B, 16.6 s and a 15.55 GB runtime cache entry for 4B in the zoo's runs. The kind is new: kits
0.1.0–0.7.3 decoderowDecisionasunknownand leave both entries out ofavailable(_:)(each tag's own catalog
code, compiled and run on this catalog.json);decisionwould list them in a shipped kit'ssystemone models, and
loading one there would download the bundle for aTypedDecisionsthat cannot read it.TypedDecisionsrefuses them by
their kind before downloading anything.KitKevDeciderrefuses a bundle whose graph takes a dynamic query length
(measured in the zoo and not shipped: its memory grows).systemone serve,systemone mcpand
SystemOne.backend(catalog:)loadKitKevDeciderfor the kind, and so dodecide-cli ask,serveandparity.
CoreAI.decideandCoreAI.systemOneload throughTypedDecisions, so they refuse Kev by its kind.
On an M4 Max throughdecide-cli parity, downloaded from the catalog pin, on the zoo fixture's 155 records that carry
their text (186 questions): the author's row ids on every question, the argmax equal to the author's fp32 read-out on
every question with a top-2 margin above 0.02 (Kev-0.8B 181/181, near-ties 4/5; Kev-4B 182/182, near-ties 3/4), max
|Δp| 0.0124 and 0.0153, and the hidden rows and p bit-equal to the zoo's Swift host's run of the same bundle on all 186
questions of each model, the answers on all 155 records (Kev-0.8B with the shared prefix on and off, Kev-4B on).
systemone serve --model kev-0.8bpassesconformance/check.py(22 requests and the three routes); a request with
imagesgets a 422.KevTests(the catalog entries, the request in the author's form, the refusals; with
KIT_KEV_GATE=1/KIT_KEV_GATE_4B=1, 10 Kev-0.8B and 3 Kev-4B records bit for bit against the zoo's Swift host,
direct and shared). -
clef-flash — catalog id
clef-flash(the newkind: jointDecision,format: jointHead), driven by the new
KitClefDecider(Sources/CoreAIKit/ClefFlash/): Cloudflare's decision model from Qwen3.5-9B (Apache-2.0). Every
question of a request (noul, choice, score) is read in one pass by its joint schema head, with the state as text or
JSON and at most one image, and every option gets a probability; nothing is generated. The host is the model zoo's
apps/ClefFlashlibrary (5ef2247) with every numeric path unchanged: the request rendered as the author's
encode_record()(canonical JSON, each piece tokenized alone), the image resized by Pillow's own integer bicubic to a
fixed grid (g448by default, 196 image tokens;g256, 64) and read by that grid's tower, the decoder (fp16, no
vocabulary head, 64 tokens per call) returning the hidden state at every position, the head graph over span means, the
last token and each option's rows of a 2.03 GB fp16 lm_head table on the host, and a float32 softmax per question at
temperature 1. Up to 16 questions and 128 options a request. 18,178 MB: the decoder, the head and the table; a tower
(909 MBg256, 912 MBg448) downloads the first time an image asks for its grid. macOS only (the decoder alone is
15.9 GB); the zoo measured a first load that specializes the decoder in 56 s into 31.8 GB of the runtime's cache, and
4.5 s from that cache. The entry names the head, the table and the towers in a newassetsfield, which a kit built
before it ignores. Its kind is the newjointDecision(a kit built before it decodes the entry asunknownand
leaves it out ofavailable(_:);decisionwould list it in a shipped kit'ssystemone models, and loading it
there would download the 15.9 GB decoder for aTypedDecisionsthat cannot read it).TypedDecisionsrefuses it by
its kind, and adecisionentry that saysjointHeadby its format, before downloading anything./v1/systemone
gainsimages(one image: a data URL, base64, or a file path, whichSystemOneServerreads only on 127.0.0.1) and
grid.systemone serve --model clef-flashand the MCP server load it through the newSystemOne.backend(catalog:);
decide-cli askanddecide-cli serveloadKitClefDeciderwhen the entry's kind isjointDecision. A backend that
reads no images answers an image request with a 422 that names it.
On an M4 Max throughdecide-cli parity, downloaded from the catalog pin: 240 of the zoo fixture's runs (200 on its
186 records, 40 on the 30 held out; its 14nativeruns, at the processor's own grid, have no tower here — the zoo's
gate fed 13 of them the author's image rows,photo_01's 1,200 image rows being more than the decoder's 1,024-row
buffer), ids and spans equal to the author's on every run, the argmax equal to the author's fp32 read-out on every
question with a top-2 margin above 0.02 (352/352 and 186/186; near-ties 3/3 and 2/2), max |Δp| 0.0122 and 0.0082, and
logits, probabilities, the decoder's hidden rows and the responses bit-equal to the zoo's Swift run of the same files
on all 240.systemone serve --model clef-flashpassesconformance/check.py(22 requests and the three routes), and
the fixture's receipt sent inimages, as a data URL and as a path, answers with the zoo Swift run's probabilities to
the wire's four decimals (14/14 values).ClefFlashTests(the catalog entry, the wire, the request in the author's
form; withKIT_CLEFFLASH_GATE=1, twelve fixture runs bit for bit against the zoo's Swift run). -
decider-2b-vision — catalog id
decider-2b-vision(the newkind: visionDecision), driven by the new
KitVisionDecider(Sources/CoreAIKit/DeciderVision/) andCoreAI.decide(image:_:_:grid:options:)(one question,
a dictionary or an array): typed decisions about an image — Mapika's decider transplanted into Qwen3.5-2B's
vision-language model (Apache-2.0). Choice, score and noul questions about an image and a state, every question of a
call read in one pass at its own answer slot, up to 10 options. The host is the model zoo'sapps/DeciderVision
library: the image resized in Pillow's pass order to a fixed grid (g256, 64 image tokens;g448, 196), a vision
tower per grid (fp16 weights, fp32 math) and a two-function decoder (int8 with three fp16 layers; S = 1main, S = 16
prefill) with the tower's rows as a static input, on the low-level runtime. 3,325 MB with theg256tower; the
g448tower (663 MB) downloads when first asked for.Decision.TiminggainsimageSecondsanddecoderSeconds
(nil for the text models). On an M4 Max throughdecide-cli parity, downloaded from the catalog pin: 108/108 slots
argmax-identical to the author's fp32 read-out, max |Δp| 0.0103, letter logits bit-equal to the zoo's Swift run;
the zoo gated the same files on an iPhone 18 Pro (device JIT, 108/108; the app needs the increased-memory-limit
entitlement), the kit has not run on a phone. Not behind/v1/systemone: the wire form has no image field.
DeciderVisionTests(the fixture's 76 prompt rows against the author's ids, with the tokenizer),
DeciderVisionSmokeTests(opt-in viaKIT_DECIDERVISION_*). -
Audio8-TTS-Preview-0.6b — catalog id
audio8-tts-preview-0.6b(kind: tts), driven by the new
Audio8TTS(Sources/CoreAIKit/Audio8TTS/): Edge0's DualAR text-to-speech (Fish Audio S2 Pro design), 11
languages, zero-shot voice cloning from a 0.5–30 s reference (Audio8Voice: the reference's codec codes +
transcript). One Core AI call per 46 ms frame: the slow AR step, the publisher's top-k / top-p / temperature
Gumbel-max draw with the RAS rule and the fast AR's ten rows run inside theframegraph; the host supplies the
uniform draws (Audio8NoiseSource, seeded by default) and streams 32-frame chunks through the causal codec.
KitSpeaker(catalog: "audio8-tts-preview-0.6b")dispatches it (44.1 kHz). Gated in the zoo on 18 fixtures against
the publisher's fp32 code (draws 1,671 / 1,695 semantic and 14,711 / 15,093 codebook, Fun-ASR round trip at the
oracle's level) and on an M4 Max in Release (about real time, RTF 1.05 measured under other load); iPhone not yet
measured.Audio8SmokeTests(opt-in viaKIT_AUDIO8_*). -
Fun-ASR-Nano-2512 — catalog id
fun-asr-nano-2512(kind: asr), driven by the new
KitFunASRModel(Sources/CoreAIKit/FunASR/): a kaldi-fbank + LFR front end, funasr's ChatML prompt
with the audio rows asvocab + slotids (hotwords, language and itn options) and the decoder's
63-rowaudio_embedsstatic input, like the Qwen3-ASR path.KitTranscriber(catalog:)dispatches
it. Chinese (dialects, Cantonese), English and Japanese; encoder fp16-weights/fp32-math 450 MB +
decoder int8 759 MB, both JIT.aimodels for macOS and iOS. Gated in the zoo on 155 clips against
the fp32 funasr oracle (150 exact, 5 knife-edge, 0 above the margin floor) and on an iPhone 18 Pro
(device JIT, RTF 0.075, first load 5.1 s then 0.9 s);FunASRSmokeTests(opt-in viaKIT_FUNASR_*). -
8-speaker diarization:
nemotron-3-diarization. NVIDIA's Nemotron-3-Diarization
(OpenMDW-1.1) tracks up to 8 speakers at 10 ms. It runs behind the sameKitDiarizeras
sortformer-diar-v2(4 speakers at 80 ms):KitDiarizer(catalog: "nemotron-3-diarization"),
orMeetingTranscriber(diarizer: "nemotron-3-diarization").CoreAI.transcribeMeetingand
MeetingTranscriber's default stay onsortformer-diar-v2.The download is 200 MB on either platform: the GPU graph, run in low latency (0.72 s chunks,
0.32 s look-ahead), and ahost/folder beside it with the projection, the silence row, the mel
filterbank and the Hann window. The host code is the model zoo's Swift port, copied into
Sources/CoreAIKit/Audio/NemotronDiarizer. Against transformers' fp32 reference, speaker
activity agrees on 99.9987 % of a 97.6 s clip and 100 % of a 21.5 s clip. A chunk takes
15.4–15.9 ms on an M4 Max GPU, measured with another job on the GPU, and 30.1 ms on an iPhone
17 Pro GPU.NemotronDiarizerTestschecks one speaker-cache update and one mel chunk bit for
bit against the zoo's host, with no weights and no audio.SpeakerSegmentnow carries itsframeSec, 0.08 unless given. Each diarizer reports its own
nSpkandframeSec.diarize(samples:bridgeFrames:)defaults tonil, which means 0.48 s
in the model's frames. The Sortformer path gives the same results as before: bit-identical
activity and the same turns on both clips.Examples/Meetinggains--diarizer,
--asr systemanddiarize-gate, which checks a diarizer against a reference's per-frame
probabilities. -
Zero-shot text classification:
gliner2.5-decide. fastino's GLiNER2.5-Decide (Apache-2.0)
answers any tasks with any labels, named at call time, from one forward:TextClassifier()
downloads it (ModelID.gliner25Decide),classify(_:tasks:)returns every label's probability
per task. The host follows gliner2 2.0.0's collator and decision rule;textclassify-cli --gate
checks both against the zoo's fixtures. Before this,TextClassifierloaded only a local export.
Examples/TextClassifydownloads the catalog model when no--bundleis given. The
entry's kind is the newtextClassification(a kit built before it decodes the entry as
unknownand leaves it out ofavailable(_:);decisionwould list it insystemone models,
whoseTypedDecisionscannot load it). Download: 1,756 MB on macOS and on iOS. -
Examples/TextClassifygains the inbox app. A support inbox of synthetic messages, sorted on
one tap: every message gets an intent, an urgency and a sentiment from one forward of
gliner2.5-decide, and the counts and the measured time (the median milliseconds per message,
messages per second, the total) are on screen, on iPhone and Mac.textclassify-cli --inbox <count>sorts the same inbox headless through the sameInboxSorterand prints the same
numbers. The inbox is generated from a seed (an invented shop, plain item names, first names
only).-autoplay 1 -log 1runs it unattended and writes the result as JSON, for a recording or
a phone read back overdevicectl. 1,000 messages took 40.9 s on an iPhone 18 Pro (iOS 27.0,
GPU, 39.2 ms median per message; the phone holds about 38 ms for the first 20 s of a run and
slows from there, 1.25–1.5× by 41 s over five runs) and 30.1 s on an M4 Max (30.0 ms),
2026-09-26.
Fixed
-
VibeVoice, VoxCPM and VoxCPM2 stop at the end of their KV cache. Their decode graphs hold a
fixed KV cache of 512 positions and write it at each step's position, and the kit did not stop
before the end: a VibeVoice voice with a long prompt (de-Spk0_man takes 510 positions, kr-Spk1_man
421) or a VoxCPM text whose tokens and frames came to 512 or more ran steps outside the cache, and
what came out of them was never checked.StatefulGraphModelnow reads the cache's length from
the graph's state (capacity), andstepthrowsVisionError.kvCacheFullfor aposat or past
it. VoxCPM and VoxCPM2 end the clip where the cache ends (the text's tokens and the clip's frames
take one position each, and the decode after the last frame, which nothing read, no longer runs)
and throwVoxCPMError.textTooLongfor a text of 511 tokens or more; a long text ends early, so
pass one sentence at a time. VibeVoice ends a sentence where a cache ends, so with kr-Spk1_man a
long sentence ends early, and throwsVibeVoiceError.voiceFillsCachefor de-Spk0_man, whose
prompt leaves 2 positions. The en-* voices, the default ones, stay inside the cache. Inputs that
stay inside the cache give the same output as before. -
Unlimited-OCR stops at the end of its KV cache. Its decoder holds a fixed KV cache of 2,048
positions and the image prompt takes 115 of them, and the kit did not stop before the end: a
maxTokensabove 1,933 ran decode steps outside the cache, and what came out of them was never
checked.KitDocReadernow reads the cache's length from the decoder's state and stops a read at
1,933 tokens, and a decode step at or past the end throwsVisionError.kvCacheFull. The default
maxTokensof 1,024 stays inside the cache; a read that stays inside it returns the same text
as before. -
A graph compiled for one iPhone is picked only on that iPhone. The loaders that pick their
own graph took an AOT.aimodelcover the JIT.aimodelon iOS, most of them by the name
.h18p.aimodelc, the iPhone 17 Pro's. The iPhone 18 Pro (h19p) refuses that graph
(incompatibleCompiledAssetArchitecture) and falls back to nothing, even with the JIT graph
beside it.GraphBundleis now the one rule, and the diarizers, the separator, the forecaster,
VibeVoice, VoxCPM, VoxCPM2, the Nemotron and Parakeet ASR loaders and the encoder decider use it:
the AOT graph when it was compiled for this device (AIModel.deviceArchitectureNameagainst the
<arch>of<name>.<arch>.aimodelc, or of the<function>-<arch>files inside), else the JIT
.aimodel, else the kit's error naming the architectures the bundle was built for
(KitBundleError.noGraphForDevice). The Mac follows the same rule;GraphBundleused to prefer
any.aimodelcthere too.VibeVoicePaths.inBundleDirnow throws; theaot(root:arch:)builders of
VoxCPM, VoxCPM2, PocketTTS and DotsTTS default to this device's architecture (wash18p), and
StableAudioPaths.resolve(root:aot:)names it. From the catalog, VibeVoice, VoxCPM and VoxCPM2
still load on the iPhone 17 Pro only, as in 0.7.3: their pins have not moved, andios/there
holds only that phone's graphs. On another iPhone they now throw that error. -
Four catalog models name their JIT graph on iOS.
sortformer-diar-v2,
nemotron-3-diarization,melband-roformer-vocalandtimesfm-2.5-200mpointed iOS at the
iPhone 17 Pro's AOT graph, which only that phone loads. They now name the.aimodelthe Mac
downloads, which an iPhone specializes on its first load (Nemotron-3-Diarization on an iPhone 18
Pro: 0.78 s the first time, 0.16 s after).ModelID.sortformerDiarV2names it on both
platforms. Whisper and VibeVoice keepiosas their path; at their pins it still holds the
iPhone 17 Pro's graphs (Nemotron 3.5 ASR's pin moves to the JIT graphs, below). -
The Nemotron-3-Diarization graph compiles for x86_64 again.
N3DGraphread and wrote
Float16arrays without the guard the other ten files carry
(#if !((os(macOS) || targetEnvironment(macCatalyst)) && arch(x86_64))), so a Release build of
any macOS app linking CoreAIKit failed on the Intel slice ('Float16' is unavailable in macOS).
The diarizer is new in this release, so no released version had this failure. On an Intel Mac a
float16 N3D bundle now throws the contract error instead; the float32 bundle is unaffected. -
Four catalog pins move to the JIT layouts.
timesfm-2.5-200m,vjepa2-vitl-ssv2,
nemotron-3.5-asr-streaming-0.6bandgliner2.5-decideare pinned to the Hub revisions of
2026-09-26 whoseios/holds the JIT graphs (the compiled ones moved toios-h18p/,
GLiNER2.5-Decide's toios-h19p/), so an iPhone of any generation loads them;sizeMBfollows
(GLiNER2.5-Decide on iOS 1,958 → 1,756 MB). A kit built before this release reads those four
with its old rule too.vibevoice-realtime-0.5bandvoxcpm-0.5bstay on their previous
revisions: the old loaders openios/<name>.h18p.aimodelcby name, and the livecatalog.json
reaches every shipped kit, so moving those two pins would stop them on the iPhone 17 Pro. They
move in a later catalog update (the two Hub repos already carry the JIT layout,b4480f87and
38fc0aea); from then an app built with 0.7.3 or earlier no longer loads these two on the iPhone
17 Pro, while 0.8.0's loaders take the JIT graphs on any iPhone.voxcpm2-2balso stays on its
previous revision (e03d15d1). -
ModelID.gliner2PIIis pinned. Itsios/carried the iPhone 17 Pro's compiled files beside
the JIT graph, which the iPhone 18 Pro refuses; since 2026-09-26 it holds the JIT graph alone
(627 MB). The preset readmain, and an app that had downloaded the oldios/kept it, since
the cache is keyed by revision; the pin (74fb5c1) moves the cache path. -
InformationExtractoropens a graph compiled for this device. It looked for a.aimodel
alone, and GLiNER2-PII'sios-h18p/at that pin holds the iPhone 17 Pro's.h18p.aimodelcand
no.aimodel, so the folder could not be opened. A*.<arch>.aimodelccompiled for this device
now wins over the.aimodel, as inGraphBundle; a folder with the JIT graph alone loads as
before. -
An iPhone takes the graphs its repository compiled for it. A model whose path is
ios(the
catalog's iOS variants, and the presets that leavepathnil) always downloadedios/, so an
ios-<arch>/folder, the same bundle compiled for one iPhone generation, was reached only by a
caller who wrote its path.ModelStorenow listsios-<arch>/first,<arch>being
AIModel.deviceArchitectureName(h18pon an iPhone 17 Pro,h19pon an 18 Pro), and takes it
when the Hub has files there; on a 404 or an empty tree it takesios/, one listing more. Any
other error of that listing is thrown, not read as absence. A copy already on disk is used,
whichever folder it is, and the Hub is not asked, so anios/downloaded before stays in use;
localURL(for:),delete(_:)and the offline fallback look inios-<arch>/first.
remoteSize(of:)anddownloadPlan(for:)follow the folder a download takes; the catalog's
sizeMBstays theios/figure, and anios-<arch>/can be larger. At the current pins an
iPhone 17 Pro takesios-h18p/forvjepa2-vitl-ssv2,nemotron-3.5-asr-streaming-0.6band
ModelID.gliner2PII, and an iPhone 18 Pro takesios-h19p/forgliner2.5-decide. Every other
path, and the Mac, is unchanged. On iOSCoreAIKitCorenow imports Core AI, for the
architecture's name alone.
Known issues
- On an iPhone 18 Pro with iOS 27.2 (build 24B5099f), Qwen3 0.6B (
qwen3-0.6b) answers with
unreadable text from its first token, and ChatDemo's entry check fails. 0.7.3 gives the same
text, so this is not new in 0.8.0. The cause is not known. Only this phone and OS build were
tried. The phone compiles the model's iPhone build (ios/) the first time it loads; not checked
yet: a build compiled ahead of time for the iPhone 18 Pro, a forced recompile on the phone,
other iPhones, other iOS builds, and running on the GPU only. On a Mac with macOS 27.0 the
model answers correctly.