Repository navigation
Releases
v0.11.0
Basalt v0.11.0
brew tap goodmai/basalt https://github.com/goodmai/basalt
brew trust goodmai/basalt
brew install goodmai/basalt/basaltBuilds from source: Xcode 27.0+ (Swift 6.4) and macOS 27+, Apple Silicon.
Source tarball SHA256: 48e35f3809acfb3afafbeed0381335a4dc661abbb7c2690ced83ddc556117244
What's Changed
- chore: point formula at v0.10.0 by @github-actions[bot] in #382
- speech: Nemotron 3.5 Transcribe (nemotron_transcribe) model, /v1/audio/transcriptions, and SpeechAnalyzer fallback (#230, #222) by @goodmai in #383
- feat(cloud): segment-level speculative cascades between local and cloud models by @goodmai in #389
- feat(embeddings): bidirectional mask, float32 mean pooling, and headless LoRA (#384) by @goodmai in #390
- feat(lab): GoodViewCollator, JEPAProjector, LeJEPALoss (#385) by @goodmai in #391
- lab: train_jepa (#386), Orthogonal Procrustes (#387), matmul Quest block scoring by @goodmai in #392
- quantization: NF4 phase (c) — native nf4 mode from our mlx fork, experts included, experimental (#357) by @goodmai in #395
- lab: distill — teacher → student LoRA from an API-key teacher, completion-only loss, cached and budgeted (#393) by @goodmai in #396
- Batch 13: fused Quest kernels, MTP draft head, abliteration and GPTQ measurements, distillation fixes (#280 #209 #44 #376) by @goodmai in #398
- Batch 14: on-policy distillation, GPU/ANE contention rule, Core AI embeddings on the ANE, prefill-split spike step 1 (#394 #273 #223 #274) by @goodmai in #401
- Batch 15: System One decisions, OCR phase 0, Apple Vision OCR, combine pipeline phase 1, compute plan (#234 #400 #403 #402 #399) by @goodmai in #406
- Batch 16: PaddleOCR-VL, preemptible distill, ANE System One, pipeline audio, NF4 kernels, MLX↔Core AI hand-off (#400 #399 #234 #402 #357 #274) by @goodmai in #411
- Batch 17: DeepSeek-OCR, pipeline presets and sequential memory, sidecars by rule, context-scaling bench, per-token trace, cascade System One (#400 #402 #399 #242 #405 #407) by @goodmai in #414
- vision: ANE vision tower for PaddleOCR-VL (einsum export, CER 1.74% vs 1.90%) behind an experimental flag; plan notes from batch 17 measurements (#224 #399 #402) by @goodmai in #415
- ocr: DeepSeek-OCR no-repeat 20-gram processor, as in the official inference; gundam CER 15.6% → 3.8% (#400 follow-up) by @goodmai in #416
- compose: basalt compose plan|up — YAML model systems with local and cloud models, on-demand/sequential keep (#412 phase 1) by @goodmai in #417
- compose: answer routes (MoM), multi-cloud decide cascades, cloud budgets and the 401/429 policy (#412 phase 2) by @goodmai in #418
- compose: named pipelines, up-as-diff on a running system, compose ps/down, docs/compose.md (#412 phase 3) by @goodmai in #419
- deps: bump github.com/apple/swift-log from 1.16.2 to 1.17.0 by @dependabot[bot] in #420
- deps: bump github.com/apple/swift-nio from 2.104.0 to 2.105.0 by @dependabot[bot] in #421
- gguf: native GSQ-RCO i-quant/k-quant kernels and GGUF import, experimental (#408 phases 0 and 2) by @goodmai in #424
- compose v2 phase 4a: schema as data, exact lines, version 2 variables and profiles, paths next to the file (#423) by @goodmai in #425
- deps: bump swift-jinja from 2.5.2 to 3.1.0 (major), checked against real chat templates by @goodmai in #426
- compose v2 phase 4b-1: cloud budgets and key variables apply to a running system (#423) by @goodmai in #428
- compose v2 phase 4b-2: admission on apply, unload-first when old and new do not fit together, plan counts every pipeline (#423) by @goodmai in #429
- compose v2 phase 4b-3: the local decision model is loaded once, by serve's decision engine (#423) by @goodmai in #430
- gguf: half-precision lattice path for IQ3_S and IQ3_XXS matvec, decode +4.7% on Qwen3.8-27B (#408) by @goodmai in #431
- compose v2 phase 4b-4: the runtime runs on narrow injected services; a re-keyed cloud model uses its new key (#423) by @goodmai in #432
- gguf: IQ4_XS matvec reads each 16-byte group with one load, −14% on the kernel (#408) by @goodmai in #433
- Mac app phase 1: SwiftUI chat on a supervised basalt helper, SwiftData history, Monitor window, bundle script (#434, closes #55) by @goodmai in #435
- compose v2 4b-5: compose up changes the chat model and the local decider without a restart; /v1/systemone follows (#423 A2) by @goodmai in #436
- compose v2 4b-6: each pipeline decides with its own local model, loaded once (#423 A1) by @goodmai in #437
- compose v2 4c-1: closed model states, /v1/health/live and /ready, unset keys fail up, pid reuse, down fails on timeout (#423) by @goodmai in #438
- compose v2 4c-2: bind first and load after; ps shows loading; a second up waits and applies; up --wait (#423) by @goodmai in #439
- compose v2 4c-3: startup probes; a failed probe shows as failed and keeps ready at 503 (#423) by @goodmai in #440
- Mac app phase 2: menu bar item, quick ask on ⌥Space, Settings with Keychain cloud keys and open at login, restored selection and drafts (#434) by @goodmai in #441
- Mac app phase 3a: Models window — fit verdicts gate loading, resumable downloads, safe delete (#434) by @goodmai in #442
- Mac app phase 3b: PDF, image and audio attachments through /v1/pipeline, perceived text kept for follow-ups (#434) by @goodmai in #443
- Mac app phase 3c: inspector — per-conversation system prompt, receipt details, perceived attachments (#434) by @goodmai in #444
- Mac app phase 4a: App Intents, Ask Basalt service, Dock drop, answer notifications (#434) by @goodmai in #445
- Mac app phase 4b: model or compose system with switching, adoption by arguments, compose and cloud spend in the Monitor, cloud mark (#434) by @goodmai in #446
- compose v2 4d: compose pull, plan sized from the Hub or refused, compose diff dry run, progress events (#423) by @goodmai in #447
- decisions: Gemini Decision 1 as a cloud stage — cloud:gemini/, multimodal media on /v1/systemone, compose data: media opt-in by @goodmai in #449
- Mac app phase 5a: VoiceOver, Russian localization with a coverage test, daily update check (#434) by @goodmai in #456
- Mac app phase 5b: zip, dmg, notarization, release step and Homebrew cask when signing secrets exist (#434) by @goodmai in #457
- telemetry: trailing windows, decision aggregates and OpenMetrics at /api/v1/metrics (#450 phase 1) by @goodmai in #458
- ci: cap Swift Testing parallelism so MLX tests on the eval lock stop starving live-server tests (#427) by @goodmai in #461
- compose v2 4e-1: cascades in any local/cloud order, readout calibration from the bundle, OpenJev sizing (#423) by @goodmai in #460
- telemetry: MoE routing and expert cache, per-token confidence shape, telemetry_query MCP tool, Monitor sections (#450 phases 2–4) by @goodmai in #459
- compose v2 4e-2: Jev LoRA applied unmerged on the stored base (4-bit included), Qwen3.5 text bases, sizing as base plus adapter (#423 J2 J3) by @goodmai in #462
- app: measure the #434 budgets — launch, idle wakeups, memory, 60 fps streaming, quick ask latency by @goodmai in #463
- jev: real-model gate for the unmerged LoRA — OpenJev-0.6B on bf16, 8-bit and 4-bit bases (#423 J2) by @goodmai in #464
- jev: a 9B judge beside Gemma 4 12B on 24 GB — 4-bit fits, 8-bit swaps (#423 4e gate, #234) by @goodmai in #465
- nf4: fair comparison from bf16 on Gemma 4 12B — NO-GO on quality (perplexity +29%), the kernel is correct (#357) by @goodmai in #466
- ane: #274 step 3a — the ANE weight format for the prefill FFN split on Gemma 4 12B: int8 per channel is the only one accurate and fast by @goodmai in #467
- ane: #274 step 3b-2 — SplitFeedForward: a dense layer's MLP as the GPU's columns plus a Neural Engine share, through Gemma 4's async prefill by @goodmai in #468
- docs: lead with what is measured — golden-tested models on the first screen, every number linked, a README model check in CI (#404) by @goodmai in h...
v0.10.0
Basalt v0.10.0
brew tap goodmai/basalt https://github.com/goodmai/basalt
brew trust goodmai/basalt
brew install goodmai/basalt/basaltBuilds from source: Xcode 27.0+ (Swift 6.4) and macOS 27+, Apple Silicon.
Source tarball SHA256: 075368acd37a181321b6ad64a5ffd6ce532495be818e56b79433049b274a28ca
What's Changed
- chore: point formula at v0.9.0 by @github-actions[bot] in #268
- feat(#255): batched decode for recurrent hybrids; fix concurrent forwards racing compiled traces by @goodmai in #269
- feat(#37): head-level KV split — retrieval heads keep the context, streaming heads keep sinks and a window by @goodmai in #271
- feat(#33): /v1/models/preview, wired limit by default, mlx-swift 0.33, watchdog for long prefills by @goodmai in #276
- fix(inference): every registerAliases caller waits until the aliases are registered by @goodmai in #279
- fix(#277): TTFT counts the prefill the iterator runs when it is built by @goodmai in #278
- feat(#33): experimental Quest block-sparse decode attention behind serve --quest-top-k by @goodmai in #281
- chore(#236): remove the GemCLI, GemLogger and GemLogLevel typealiases by @goodmai in #283
- docs(#284): size and complexity limits with a ratchet; Swift 6.4 adoption in CONTRIBUTING and the skills by @goodmai in #285
- docs: register Xcode's MCP server and route API questions to the xcode-mcp skill by @goodmai in #288
- test: rate-limit router test no longer depends on runner speed by @goodmai in #287
- refactor(#284): JSONSchemaGate — split checkDroppedKeywords and visit along their seams by @goodmai in #289
- ci(#284): size and complexity ratchet on changed Swift files by @goodmai in #286
- refactor(#284): DiffParser.parseUnifiedDiff — one flush, body lines classified in diffLine by @goodmai in #293
- refactor(#284): ServeCommand.run and ModelsCommand.run — one memory report, run split along its seams by @goodmai in #295
- feat(#270): choose KV heads by ablation; measure head shares on harder tasks than passkey by @goodmai in #296
- ci(#284): type ≤ 200 / extension ≤ 150 lines in the ratchet; value core, actor shell in CONTRIBUTING by @goodmai in #298
- ci(#284): comment checker and xcode-mcp helper in Swift by @goodmai in #299
- docs(#284, #300): legacy types ≤ 450, race-free by construction rules; size checker closes its pipes by @goodmai in #302
- ci: Unit Tests skip docs-, skills- and checker-only changes on the single Mac runner by @goodmai in #306
- test(#250): mark the Nemotron-H prompt-lookup equivalence as a known intermittent issue by @goodmai in #307
- refactor(#284): MLXInferenceEngine.launch — FreshCache plan as a value core, one function per path by @goodmai in #303
- types(#304): TokenCount and ByteCount units in KVCacheGeometry's public API by @goodmai in #305
- refactor(#284): MLXInferenceEngine — load split along its seams, static helpers out of the actor body by @goodmai in #308
- ci: Swift checker jobs run in the swift:6.1-noble container (Swift is leaving the Ubuntu runner images) by @goodmai in #312
- refactor(#284): MCPServer — value core for routing, arguments and results; 677 → 40 lines by @goodmai in #310
- refactor(#284): ServeCommand — option groups resolve into Sendable values (692 → 186 lines) by @goodmai in #309
- refactor(#284): SessionLogWriter — value core, actor shell by @goodmai in #311
- refactor(#284, #300): value cores for controllers, router, orchestrator, hub, disk cache, fit/inspector, schema gate, harness, table renderer; CacheBlendStore and MTPPrefix without @unchecked by @goodmai in #313
- refactor(#284, #300, #33): function limits across CLI, server, core, harness and inference; prompt-cache leases stay on the engine actor; MTP drafter inputs as values; Quest review fixes by @goodmai in #314
- concurrency(#300, #304): zero @unchecked Sendable — prompt caches, MTP drafter and batch state owned by the model context; typed prompt-cache bytes by @goodmai in #320
- refactor(#300, #315–#318, #321): generation plan, engine state machine, one MLX owner, deterministic tests, property tests, CI ratchets, exhaustive error mapping by @goodmai in #330
- feat(#322–#327, #300): capability registry, metamorphic matrix, reference logits, parity runner, load-time probe, health monitors, fault injection by @goodmai in #331
- feat(#328): model X-ray — MCP tools for structure, tensors, KV cache, traces (logit lens, attention, experts) and sandboxed experiments (steering, ablation, weight deltas, compare) by @goodmai in #333
- refactor(#284): tail — engine stages, relay, DTO inits, Task.detached gone; #323 KV quantization gate, #326 probe ceilings, #321, #301 by @goodmai in #334
- fix(#329): Nemotron-H group routing takes the two largest scores per group; lfm2_moe reference uses transformers' router by @goodmai in #335
- feat(#332): research modes over MCP — gate, live adapters, pack/unpack, introspection, contrastive steering, self-training, weight conversion by @goodmai in #336
- test(#319): batched decode contract — quality parity with solo, exact invariants pinned on tiny and real hybrids by @goodmai in #337
- fix(#325): health monitors checked on real models — a reply cut off at max_tokens is not degenerate; tiny-model tests; mid-stream error message by @goodmai in #338
- fix(#250): the Mamba-2 decode kernel wrote no state for state sizes below 32 — the lookup flake's root cause; pin cd871d3 by @goodmai in #339
- refactor(#304): KV and memory budgets are ByteCount end to end; decode tok/s unchanged by @goodmai in #340
- test(#322): LFM2.5-24B parity goldens — the benchmark prints the reasoning when the answer is empty by @goodmai in #342
- refactor(#284): exhaustive switches over basalt's own enums — the default: audit by @goodmai in #341
- refactor(#300): stage 3 — reducer transitions in the orchestrator, router, hub and session-log actors; three reentrancy races fixed by @goodmai in #343
- test(#291): real-model regression gate — golden top-64 logit signatures (KL 1e-4) on Qwen3.8-27B, LFM2.5-24B and Gemma 4 26B-A4B, plus a footprint check by @goodmai in #347
- feat(#346): regex, EBNF and structural-tag response formats; fix constrained answers reported as reasoning on LFM2.5 by @goodmai in #348
- feat(#345): image input end to end on REST and MCP — images were dropped; Gemma 4 26B-A4B answers the same through all three paths by @goodmai in #349
- feat(#346): constrained requests join the batched decoder, each row with its own grammar mask by @goodmai in #350
- Batch: Gemma 4 joins the batched decoder (#255), embeddings yield to chat (#233), roofline with prefill (#272), seeded MTP tests (#290) by @goodmai in #354
- Batch: download digests and a symlink-blob fix (#353), rerank yields to chat (#355), --kv-bits refusal (#165), jump-forward with prefill (#205) by @goodmai in #356
- Batch: prompt lookup with quantized KV, 2.1–2.4× on code (#163), thermal pacing (#45), chat history digest (#23) by @goodmai in #368
- Batch 8: YAML config, LoRA adapters, inspect --verify, basalt review, release CI, experimental zero-copy body (#344 #179 #367 #35 #206 #345) by @goodmai in #369
- Batch 9: acceptance gate, verify overhead, small-M matmul, spectral and key-trajectory spikes, sessions LTL check (#246 #210 #208 #358 #364 #359) by @goodmai in #375
- ci: bump github/codeql-action/analyze from 4.39.1 to 4.39.2 by @dependabot[bot] in #371
- deps: bump github.com/hummingbird-project/hummingbird from 2.28.0 to 2.28.1 by @dependabot[bot] in #373
- ci: bump github/codeql-action/init from 4.39.1 to 4.39.2 by @dependabot[bot] in #370
- deps: swift-transformers 1.4.0 with compatibility for its breaking changes (tokenizer errors, BPE decoding) by @goodmai in #378
- NF4 phases (a) and (b) behind experimental opt-ins: codec, tensor_pack, convert, fork loader fix, error/KL/speed runners (#357) by @goodmai in https://...
v0.9.0
Basalt v0.9.0
brew tap goodmai/basalt https://github.com/goodmai/basalt
brew trust goodmai/basalt
brew install goodmai/basalt/basaltBuilds from source: Xcode 27.0+ (Swift 6.4) and macOS 27+, Apple Silicon.
Source tarball SHA256: f2b314c4e8f3c7d33a9c699667bf2eb2f046af9a52f03a395037f850a19024ec
What's Changed
- chore: point formula at v0.8.0 by @github-actions[bot] in #212
- fix(speculative): self-review of the 0.8.0 MTP path (#204) by @goodmai in #211
- fix(#213): untrusted Hub names, clipboard deadlock, banned primitives by @goodmai in #215
- test(#213): drop sleep-then-assert from MCP and telemetry tests by @goodmai in #216
- refactor(#213): hygiene — force-unwraps, Gem leftovers, stale comments, onboarding out of Core by @goodmai in #217
- fix(inference): finish the prefill on its own thread before the decode worker takes over (#214) by @goodmai in #218
- feat(#177): DRY and XTC samplers, experimental behind serve --experimental-samplers by @goodmai in #231
- feat(#178): /v1/embeddings and /v1/rerank on MLXEmbedders / MLXRerankers by @goodmai in #232
- fix(#213): try? that silently lost data; ProgressBar takes a clock by @goodmai in #235
- docs(readme): 0.8 measurements, features at a glance, client setup (EN + RU) by @goodmai in #239
- feat(#219): recurrent anchors at message boundaries on hybrid models by @goodmai in #244
- fix(#238): flaky tests under load — deterministic drain check, hang-guard headroom, xcode-toolchain RC race by @goodmai in #248
- feat(#88): X-Basalt-Cache-Key opens a client prompt-cache namespace by @goodmai in #249
- feat(#220): MemoryCoordinator — react to macOS memory pressure at runtime by @goodmai in #252
- fix(#251): bump mlx-swift-lm to 1830ae4f — hybrid models no longer leak ~0.65–1.7 GB per load by @goodmai in #253
- feat(#240): batched decode for plain-attention models behind serve --batch-slots by @goodmai in #254
- docs(#247): Gemma 4 prompt-cache reuse is exact up to chunking drift by @goodmai in #256
- feat(#241): stricter lookup backoff on hybrids; speculative bench at T=0.7 by @goodmai in #257
- feat(#227): models check --deep refuses a broken repo from its headers, before the download by @goodmai in #258
- fix(#241): drop the hybrid-only acceptance threshold now that replay is fused by @goodmai in #259
- feat(#176): outcome-driven escalation bias for the Bradley–Terry router by @goodmai in #260
- feat(#207): MTP blocks deeper than 2 on Qwen3.5-family targets (--mtp-block) by @goodmai in #261
- test(#228): shared tiny-model fixtures for every family, plus a kernel admission rule by @goodmai in #262
- feat(#241): adaptive MTP draft depth by default, draft acceptance in /api/v1/stats by @goodmai in #263
- feat(#221): fused MoE router top-k for LFM2 MoE and Gemma 4 decode by @goodmai in #264
- feat(#161): prompt lookup on sliding-window models (Gemma 4) via staged KV rounds by @goodmai in #265
- feat(#207): MTP drafting with prompt-cache reuse by @goodmai in #266
- release: 0.9.0 — deeper adaptive MTP with prompt-cache reuse, lookup on Gemma 4, fused MoE router by @goodmai in #267
Full Changelog: v0.8.0...v0.9.0
v0.8.0
Basalt v0.8.0
brew tap goodmai/basalt https://github.com/goodmai/basalt
brew trust goodmai/basalt
brew install goodmai/basalt/basaltBuilds from source: Xcode 27.0+ (Swift 6.4) and macOS 27+, Apple Silicon.
Source tarball SHA256: d92a7736971da20ef9d5961b92c07cad91e1e0d7e6cee8f488ee4f422b58008b
What's Changed
- chore: point formula at v0.7.0 by @github-actions[bot] in #169
- feat(kv): experimental L2-norm key eviction in the zoned cache (#43) by @goodmai in #170
- feat(speculative): opt-in Token Recycling chain drafts for prompt lookup (#38) by @goodmai in #173
- feat(speculative): prompt lookup on dense recurrent hybrids via snapshot and replay (#162) by @goodmai in #180
- Persist the prompt cache to disk so a resumed session skips the prefill; basalt chat reuses KV between turns (#174) by @goodmai in #181
- Reseed the Nemotron-H lookup fixture off a near-tie and pin the margin (#162, #180) by @goodmai in #183
- Continue a trailing assistant message instead of answering it in a new turn (#98) by @goodmai in #182
- docs(contributing): Swift 6.4, types first, performance, smallest change by @goodmai in #184
- chore(toolchain): move to Swift 6.4 (Xcode 27) (#185) by @goodmai in #186
- fix(serve): shield shutdown from cancellation and drain MCP-only serve (#187) by @goodmai in #191
- fix(inference): ~Copyable single-owner guards for PromptCacheLease, flock, and GPU lock (#188) by @goodmai in #192
- test: Swift Testing 6.4 repetition gate, CustomTestReflectable token diffs, and drop XCTest (#189) by @goodmai in #193
- fix(inference): safe RawSpan safetensors header parsing and zero per-token scratch allocation (#190) by @goodmai in #194
- ci: pin GitHub Actions to full SHAs, reuse main test results, header-aware metallib freshness (#63) by @goodmai in #195
- fix(rest): provider-accurate error envelopes, Retry-After headers, and pre-stream failures (#76) by @goodmai in #196
- feat(ratelimit): per-client token buckets, separate burst size, and bounded LRU eviction (#62) by @goodmai in #197
- refactor(arch): finish Basalt* split, drop Sources/mlx, migrate Gem names (#64) by @goodmai in #198
- refactor(ui): drop Splash dependency and eliminate AppKit linkage from CLI (#65) by @goodmai in #199
- perf(lookup): back off and adapt draft limit on low-yield drafts (#172) by @goodmai in #200
- [P2] Grammar jump-forward for constrained decoding (#175) by @goodmai in #201
- fix(audit): 59-point post-v0.7 audit, token recycling, forced spans, and file modes by @goodmai in #202
- release: 0.8.0 — MTPLX compatibility (Bonsai 2, --mtp), version bump, CHANGELOG by @goodmai in #203
Full Changelog: v0.7.0...v0.8.0
v0.7.0
Basalt v0.7.0
brew tap goodmai/basalt https://github.com/goodmai/basalt
brew install goodmai/basalt/basaltBuilds from source: Xcode 26.4+ (Swift 6.3) and a macOS it runs on, Apple Silicon.
Source tarball SHA256: 16c7c2c9c3416167a3e38813973542dd911a13b143d00fc9e32148bb62ed5b1e
What's Changed
- chore: point formula at v0.6.0 by @github-actions[bot] in #155
- docs: real-model runs go through Swift runners by @goodmai in #156
- docs(skill): GPU lock with an owner, released once; speed from release builds by @goodmai in #157
- feat(inference): per-request speculative_lookup; back off drafting after rejections (#138) by @goodmai in #158
- feat(memory): --kv-zones, fp16 sinks and recent window around a quantized middle (#139) by @goodmai in #159
- feat(cache): CacheBlend recomputes chunk heads first; default ratio 0.3 (#137) by @goodmai in #164
- feat(memory): --kv-bits k8v4, 8-bit keys and 4-bit values in the zoned middle (#24) by @goodmai in #166
- feat(lookup): T > 0 exactness guards, draft limit 16, speculative_lookup_bench runner (#25) by @goodmai in #167
- release: 0.7.0 — version bump, CHANGELOG by @goodmai in #168
Full Changelog: v0.6.0...v0.7.0
v0.6.0
Basalt v0.6.0
brew tap goodmai/basalt https://github.com/goodmai/basalt
brew install goodmai/basalt/basaltBuilds from source: Xcode 26.4+ (Swift 6.3) and a macOS it runs on, Apple Silicon.
Source tarball SHA256: 3fe4e945c58b11688c379e1e929d3391b3b3f16b561f42324fb7e5419ef785c2
What's Changed
- chore: point formula at v0.5.0 by @github-actions[bot] in #144
- test(cache): perplexity/TTFT for CacheBlend and a twelve-turn hybrid drift session (#137) by @goodmai in #145
- feat(inference): prompt-lookup speculative decoding, opt-in (#138) by @goodmai in #146
- feat(sessions): session identity, resume after restart, idle suspend, basalt sessions, chat --resume (#106, part 4) by @goodmai in #147
- feat(gate): priority admission; a stream's prefill is cancellable by shutdown (#140) by @goodmai in #148
- feat(memory): --kv-bits auto, and KV room that survives the fp16 prefill (#139) by @goodmai in #149
- perf(decoding): lazy vocab index for the JSON mask, LRU grammar pool (#141) by @goodmai in #150
- fix(chat): Jinja2 whitespace control in basalt, reasoning_content alias (#104) by @goodmai in #151
- release: 0.6.0 — version bump, CHANGELOG by @goodmai in #152
- test: the flock holder's hang guard waits 300 s, not 30 s by @goodmai in #153
- ci(release): run the release tests serially; drop timing from two tests by @goodmai in #154
Full Changelog: v0.5.0...v0.6.0
v0.5.0
Basalt v0.5.0
brew tap goodmai/basalt https://github.com/goodmai/basalt
brew install goodmai/basalt/basaltBuilds from source: Xcode 26.4+ (Swift 6.3) and a macOS it runs on, Apple Silicon.
Source tarball SHA256: d4ff0047faa7caa3bfb225c718b6cd0df94e74a3981eee431889430493b705bd
What's Changed
- chore: point formula at v0.4.0 by @github-actions[bot] in #131
- refactor(api): one ChatPipeline for /v1/messages, /v1/chat/completions and /v1/responses (#106, part 1) by @goodmai in #132
- feat(serve): JSONL session log — records, writer, orchestrator hook, flags, privacy (#106, part 2) by @goodmai in #133
- feat(serve): session log redaction and prefix de-duplication (#106, part 3) by @goodmai in #134
- release: 0.5.0 — version bump, CHANGELOG by @goodmai in #135
- test: build fake secrets at runtime so the release secret scan stays clean by @goodmai in #136
- test(#74): wait for gate states instead of fixed sleeps by @goodmai in #143
New Contributors
- @github-actions[bot] made their first contribution in #131
Full Changelog: v0.4.0...v0.5.0
v0.4.0
Basalt v0.4.0
brew tap goodmai/basalt https://github.com/goodmai/basalt
brew install goodmai/basalt/basaltBuilds from source: Xcode 26.4+ (Swift 6.3) and a macOS it runs on, Apple Silicon.
Source tarball SHA256: 06a6e3d88dba0c6d60f38521ac60ecb36114effa8d8958bd5062542758b054ab
What's Changed
- chore: point formula at v0.3.0; fix the release formula push by @goodmai in #70
- feat(sampling): entropy-based dynamic temperature, per-request min_p (#40) by @goodmai in #69
- docs: CONTRIBUTING — no comments in code, enforced in CI by @goodmai in #71
- feat(cache): prefix KV-cache reuse per conversation slot (#56) by @goodmai in #77
- feat(cache): radix prefix tree for the prompt KV cache (#32) by @goodmai in #80
- feat(cache): prefix reuse for hybrid models via SSM state snapshots (#79) by @goodmai in #82
- fix(memory): one KV-cache geometry for every memory estimate (#75) by @goodmai in #84
- feat(cache): CacheBlend non-prefix KV reuse, opt-in (#36) by @goodmai in #83
- fix(security): loopback Host/Origin/Content-Type guard; REST model ids only (#72) by @goodmai in #92
- feat(api): tool calling on /v1/messages and /v1/chat/completions (#73) by @goodmai in #93
- fix(arc-agi): run hybrid-mode generated code in a sandboxed child (#78) by @goodmai in #94
- feat(harness): Harness² PoC — KV-cache reuse plan, Claude SKILL.md parser & MCP ExperienceLibrary (#90) by @goodmai in #91
- fix(api): send chat turns as messages through the native chat template (#86) by @goodmai in #97
- fix(models): run Nemotron-H Nano 4B and Nano 9B v2 out of the box (#100) by @goodmai in #101
- feat: thinking budget, model sampling defaults and reasoning_effort; ARC prompt fixes (#105) by @goodmai in #107
- fix(mcp): remove playwright_screenshot, which interpolated the url into Python (#67) by @goodmai in #108
- fix(cache): prompt-cache reuse for VLM-loaded models on text-only chats (#103) by @goodmai in #109
- feat(inference): grammar-constrained JSON decoding via XGrammar (response_format, forced tool calls) (#42) by @goodmai in #110
- fix(cloud): ModelRouter routes to the chosen local model, surfaces cloud stream errors, reads free RAM (#89) by @goodmai in #112
- feat(cloud): calibrated Bradley–Terry routing with cache stickiness, session/key budgets, deterministic fallback chain (#34) by @goodmai in #115
- fix(inference): stop verbatim loops in the generation loop; decode CacheBlend token texts outside the lock (#87) by @goodmai in #116
- fix(models): LFM2.5 MoE out of the box: thinking_budget and generation_config repetition_penalty (#113) by @goodmai in #114
- feat(hybrid): local ↔ cloud arbiter in the orchestrator, OpenAI-compatible fallback endpoint, switchable in the config (#22) by @goodmai in #117
- feat(serve): cost routing flags, x-session-id stickiness, virtual keys with budgets (#34) by @goodmai in #118
- fix(memory): max_tokens is capped by the KV room; --max-tokens is a ceiling; weights counted once (#85) by @goodmai in #119
- fix(orchestrator): bounded concurrent generations, single-flight model switch, drain on shutdown (#74) by @goodmai in #120
- fix(api): refuse a full queue (503) or an oversized prompt (413) before the SSE starts (#121) by @goodmai in #122
- fix(inference): drop the newlines that open the answer after (#123) by @goodmai in #124
- fix(memory): models check judges the fit on the Metal working set, not momentary free memory (#125) by @goodmai in #126
- fix(security): control tokens in message text cannot forge chat turns (#99) by @goodmai in #127
- fix(arc-agi): numpy transforms get the matrix, not a Grid object (#96) by @goodmai in #128
- deps: Hummingbird 2.28 (cap lifted), swift-jinja 2.5.2, swift-service-lifecycle 2.13 and transitive minor bumps by @goodmai in #129
- release: 0.4.0 — version bump, CHANGELOG by @goodmai in #130
Full Changelog: v0.3.0...v0.4.0
v0.3.0
Basalt v0.3.0
brew tap goodmai/basalt https://github.com/goodmai/basalt
brew install goodmai/basalt/basaltBuilds from source: Xcode 26.4+ (Swift 6.3) and a macOS it runs on, Apple Silicon.
Source tarball SHA256: a1c0c7285cd43b598e46d378eafb65f875b63689584d1eba5a1181efa9f54d4e
What's Changed
- ci: bump actions/checkout from 4 to 7 by @dependabot[bot] in #17
- ci: bump github/codeql-action from 3 to 4 by @dependabot[bot] in #18
- deps: bump github.com/hummingbird-project/hummingbird from 2.26.0 to 2.27.0 by @dependabot[bot] in #13
- deps: bump github.com/hummingbird-project/hummingbird from 2.27.0 to 2.27.1 by @dependabot[bot] in #47
- deps: bump github.com/huggingface/swift-transformers from 1.3.3 to 1.3.5 by @dependabot[bot] in #46
- deps: bump github.com/apple/swift-markdown from 0.8.0 to 0.9.0 by @dependabot[bot] in #12
- deps: bump github.com/onevcat/rainbow from 4.2.1 to 4.2.2 by @dependabot[bot] in #11
- feat(server): enforce mandatory api-key off loopback and non-destructive config update (#28) by @goodmai in #48
- feat(agent): implement agent onboarding, inspect/check, MCP lifecycle, and reproducibility tests (#29) by @goodmai in #49
- feat(observability): --stats flag, SwiftUI Monitor tab, and MCP telemetry (#30) by @goodmai in #50
- feat(quantization): four-dimensional scoring, MoE accounting, and empirical calibration (#31) by @goodmai in #51
- feat(api): coding agent support with /v1/responses, reasoning_effort, and service fast path (#27) by @goodmai in #52
- feat(cache): tool-call repair and CI for #26 (prefix KV reuse → #56) by @goodmai in #53
- ci: release guards, concurrency/timeouts and formula audit (stacked on #53) by @goodmai in #54
- feat(rest): token-bucket rate limit, drop // removed stubs (#19) by @goodmai in #57
- ci: run each PR once, build incrementally, fix model-tests, formula via branch by @goodmai in #58
- refactor: split into layered Basalt* targets, rename Gem → Basalt (#20) by @goodmai in #60
- chore: consolidate tests, scripts and assets; cover REST controllers (#21) by @goodmai in #61
- release: 0.3.0 — version bump, CHANGELOG, docs by @goodmai in #66
- fix(release): build on macos-26, require Xcode 26.4+ (Swift 6.3) by @goodmai in #68
Full Changelog: v0.2.0...v0.3.0
v0.2.0
Basalt v0.2.0
brew tap goodmai/basalt https://github.com/goodmai/basalt
brew install goodmai/basalt/basaltBuilds from source: Xcode 16+, Apple Silicon, macOS 15+.
Source tarball SHA256: ada24da23888a2d08f69d901fee0c839214b56e4d043cc6366cda048243dd81a
What's Changed
- deps: bump github.com/vapor/console-kit from 4.16.0 to 4.16.1 by @dependabot[bot] in #6
- deps: bump github.com/apple/swift-markdown from 0.7.3 to 0.8.0 by @dependabot[bot] in #7
- deps: bump github.com/apple/swift-argument-parser from 1.7.1 to 1.8.2 by @dependabot[bot] in #8
- docs: README pitch, 60-second demo, honest numbers by @goodmai in #14
- ci: audit job needs pull-requests:write for its PR comment by @goodmai in #16
- chore: rename gemm to basalt, add logo and CODEOWNERS by @goodmai in #15
New Contributors
- @dependabot[bot] made their first contribution in #6
- @goodmai made their first contribution in #14
Full Changelog: v0.1.0...v0.2.0