https://cybertoolchain.io/tools/vmlx New releases and features in vMLX, tracked by The Cyber Toolchain. en Sat, 15 Aug 2026 10:40:58 GMT https://github.com/jjang-ai/vmlx/releases/tag/v1.6.29 https://github.com/jjang-ai/vmlx/releases/tag/v1.6.29 Sat, 15 Aug 2026 10:40:58 GMT vMLX 1.6.29 adds hybrid prefix cache for 20x faster follow-ups and proves stable 100k-context multiturn with 99.9% cache reuse. • Hybrid prefix cache delivers up to 20x faster follow-up turns on long documents (43.7k-token document follow-ups drop from 107s to 5s, byte-identical at temperature 0). • Chunked SSM re-derive ships enabled by default, restoring long-context cache reuse from 0% to 99.9% above 12.5k tokens on hybrid SSM model families. • 100k-context multiturn proven stable: 97.6k-token conversations retrieve planted facts with 99.9% cache reuse and no memory faults. • Per-family cache, parser, and generation-default hardening verified across 12 model families, including DSV4 Flash native composite caching, Gemma 4 mixed-SWA with audio, Qwen 3.6/3.8 hybrid lines with native MTP, and TurboQuant KV bundles. Full history on Cyber Toolchain: https://cybertoolchain.io/tools/vmlx https://github.com/jjang-ai/vmlx/releases/tag/v1.6.28 https://github.com/jjang-ai/vmlx/releases/tag/v1.6.28 Fri, 14 Aug 2026 07:19:03 GMT vMLX v1.6.28 adds native Qwen3.6-27B bundle support with automatic multi-token prediction, delivering a measured +33% throughput gain. • Supports the Qwen3.6-27B bundle line (and the upcoming Qwen 3.8 line): the `qwen3_coder` tool-parser name now resolves to the XML-function parser, capability-only stamps route through the stock loader instead of the JANG codec, and the bundled multi-token prediction head is automatically constructed, quantized per the stamp's per-module overrides, and engaged at the stamp's trained speculative depth — no flags required. • Delivers a measured 23.3 → 31 tokens/sec (+33%) on the 4-bit Qwen3.6-27B bundle with greedy output byte-identical whether the MTP head is on or off, and image/video probes answering correctly through both API and app. Full history on Cyber Toolchain: https://cybertoolchain.io/tools/vmlx https://github.com/jjang-ai/vmlx/releases/tag/v1.6.27 https://github.com/jjang-ai/vmlx/releases/tag/v1.6.27 Tue, 11 Aug 2026 09:47:36 GMT vMLX 1.6.27 adds Muse Glimmer 30B with vision/video, expands prefix cache to 262,144 tokens, and wires up Anthropic `thinking.budget_tokens` and Ollama `think` string levels. • Ollama `think` string levels (`'low'`, `'medium'`, `'high'`) now engage thinking and select effort depth — previously they were silently ignored. • Prefix cache index expanded from ~64,000 to 262,144 tokens, with automatic migration of existing sessions, so long conversations no longer report zero reuse. • Adds full support for Muse Glimmer 30B: text tower, windowed vision tower, recipient-routed reasoning rail, and ATEM tool dialect — all three bundles (JANG_2D / 4M / 6M) supported. • Muse Glimmer 30B single- and multi-image prompts now correctly identify distinct objects in the same turn. • Muse Glimmer 30B video clips are now read in temporal order. • Chat Settings for Muse Glimmer 30B now exposes all four reasoning depths — Low / Medium / High / Extra High — replacing the previous Auto/On-only toggle. • Each reasoning depth in Muse Glimmer 30B maintains its own cache lineage, so switching back to a prior depth resumes that chain rather than starting over. • Paged RAM is now consistently enabled for vision models between the app and command line, eliminating a silent cache-reuse penalty in the app. • Control markup (recipient markers) is now scrubbed from visible Muse answers on malformed or truncated streams, while reasoning text and tool-call bodies are left intact. Full history on Cyber Toolchain: https://cybertoolchain.io/tools/vmlx https://github.com/jjang-ai/vmlx/releases/tag/v1.6.25 https://github.com/jjang-ai/vmlx/releases/tag/v1.6.25 Sat, 08 Aug 2026 20:40:39 GMT vMLX 1.6.25 brings a 29–58% DeepSeek V4 prefill speedup, sub-3s retry TTFT via KV rewind, and honest 413 errors instead of Metal OOM crashes on extreme context lengths. • New indexed-attention prefill kernel (on by default) lifts DeepSeek V4 Flash prompt processing from 347 to ~449 tokens/s at 15k context (+29%) and from 199 to ~316 tokens/s at 40k context (+58%), while sustaining 33+ tokens/s decode at both lengths. • KV cache rewind on abort/retry: engine rewinds the live KV cache to the shared prefix instead of re-prefilling, dropping retry time-to-first-token at 15k context from 46s to 2.3s. • Two-pass answer path now reuses the first pass's KV cache, eliminating a full prompt re-prefill on the second pass. • KV pool quantization keeps cache RAM under 8 GB even near the model's maximum context. • Engine now advertises the true hardware memory ceiling at startup, shrinks prefill chunks adaptively under Metal memory pressure, and returns a clean 413 on over-ceiling requests instead of crashing with a Metal OOM; the flat request timeout that killed healthy very-long prefills is removed. Full history on Cyber Toolchain: https://cybertoolchain.io/tools/vmlx https://github.com/jjang-ai/vmlx/releases/tag/v1.6.22 https://github.com/jjang-ai/vmlx/releases/tag/v1.6.22 Mon, 03 Aug 2026 05:54:42 GMT vMLX 1.6.22 scales the MLX cache ceiling with installed memory, delivering ~11% faster decoding for DeepSeek V4 Flash. • Scales the MLX cache ceiling dynamically with installed memory (16 GB → 4 GB, 64 GB → 11.5 GB, 128 GB → 23 GB, capped at 24 GB), enabling ~11% faster decode throughput for DeepSeek V4 Flash at every context length by eliminating redundant buffer re-allocation. Full history on Cyber Toolchain: https://cybertoolchain.io/tools/vmlx https://github.com/jjang-ai/vmlx/releases/tag/v1.6.20 https://github.com/jjang-ai/vmlx/releases/tag/v1.6.20 Sun, 02 Aug 2026 12:31:57 GMT vMLX v1.6.20 adds native DeepSeek V4 Flash runtime support with DSML tooling, pool-quantization, and enhanced cache reporting. • Adds native DeepSeek V4 Flash (DSV4) session support with native reasoning, DSML tool support, sampling guidance, cache-tier capabilities, and pool-quantization behavior. • Server and Chat settings now distinguish native DSV4 compiled decode and pooled-cache state from generic TurboQuant and unsupported whole-model cache modes, with explicit Activation-QAT control and effective-state reporting. • Native DSV4 cache handling preserves short append checkpoints, lossless L2 writes, valid eviction ancestry, partial-tail replay, SSD-only operation, and RAM-to-SSD refault. • Health output now distinguishes the served model from the currently loaded model. • Explicit visible-final instructions after tool use now enter a no-more-tools continuation pass, with terminal abort cleanup drained before the runtime returns to idle. • Electron settings warn when a DSV4 top-p value differs from bundle guidance and preserve remote model tool contracts. • Release packaging now attests the native Python DSV4 encoder, nested affine defaults, mixed module quantization metadata, and the clean JANG runtime source, with separate Sequoia and Tahoe packaging/notarization contracts. Full history on Cyber Toolchain: https://cybertoolchain.io/tools/vmlx https://github.com/jjang-ai/vmlx/releases/tag/v1.6.19 https://github.com/jjang-ai/vmlx/releases/tag/v1.6.19 Fri, 31 Jul 2026 16:15:27 GMT vMLX 1.6.19 adds native-MTP diagnostics, model-owned sampling defaults, separate macOS Tahoe builds, and scoped sparse-cache blocks for MiniMax M3. • Adds native-MTP health and profiling telemetry exposing bounded cache-lifecycle, acceptance, and phase-timing metrics for Qwen MTP models. • Introduces model-owned generation defaults: temperature, Top P, Top K, repetition penalty, and output limits now hydrate from the selected model bundle in new and restored chats. • Ollama requests now preserve an explicitly supplied `top_k` value, including `0`, through request translation. • Automatic prompt limits are now derived from the actually loaded model's configuration, bounded by its declared context ceiling and current memory budget. • MiniMax M3 native sparse-cache blocks are scoped to the prompt-prefill shape, preventing numerically incorrect Lightning Indexer state restoration across mismatched prefill shapes. • Cache namespaces for looped-transformer bundles now include the effective repeated-layer layout, preventing cross-incompatible-loop-count cache reuse. • Compatible Nanbeige looped-transformer JANG bundles now validate loop count, effective cache-slot count, and runtime metadata before generation, rejecting inconsistent bundles. • MiniMax M2-family automatic cache storage defaults to the q8 correctness-first policy. • Persistent cache payload encoding is moved off the inference-completion path, reducing synchronous CPU work at end of request. • KaTeX rendering gains improved disambiguation of literal currency symbols from inline math and unclosed dollar delimiters. • New settings and compatibility messages are localized across English, Spanish, Japanese, Korean, and Chinese interfaces. • Separate Apple-silicon downloads are now provided for macOS Tahoe and Sequoia-compatible systems. Full history on Cyber Toolchain: https://cybertoolchain.io/tools/vmlx https://github.com/jjang-ai/vmlx/releases/tag/v1.6.18 https://github.com/jjang-ai/vmlx/releases/tag/v1.6.18 Mon, 27 Jul 2026 19:38:42 GMT vMLX 1.6.18 adds a process-wide SSD cache budget with LRU eviction and detailed cache reporting. • Adds a process-wide SSD cache budget with coordinated LRU eviction and off-request-path publication for improved prefix reuse management. • Introduces detailed cache reporting covering attempted prefixes, applied reuse, uncached suffixes, reconstruction, fallback prefill, and aggregate disk usage. • Supports MiniMax tool arguments emitted as request-schema XML children. • Retains media across Responses tool continuations and enables local attachment playback in the Electron app. Breaking changes: • This release requires JANG 2.5.34. Full history on Cyber Toolchain: https://cybertoolchain.io/tools/vmlx https://github.com/jjang-ai/vmlx/releases/tag/v1.6.16 https://github.com/jjang-ai/vmlx/releases/tag/v1.6.16 Thu, 23 Jul 2026 03:53:28 GMT vMLX 1.6.16 ships KaTeX-backed Electron renderer for chat and reasoning surfaces and hardens bundled engine provenance. • Adds KaTeX-backed Electron renderer path for chat and reasoning surfaces. • Hardens packaged engine provenance so signed app sessions use bundled Python/vmlx_engine instead of stale development or system virtualenvs. Full history on Cyber Toolchain: https://cybertoolchain.io/tools/vmlx https://github.com/jjang-ai/vmlx/releases/tag/v1.6.7 https://github.com/jjang-ai/vmlx/releases/tag/v1.6.7 Mon, 13 Jul 2026 05:23:18 GMT vMLX 1.6.7 enables paged KV cache by default for major text model families and turns reasoning ON by default for all reasoning-capable models. • Paged KV cache now defaults ON for autodetected text families (Qwen3.5/3.6, Zaya, Nemotron-H, LFM2.5, Laguna, Hy3, Step-3.7) when continuous batching and prefix cache are active; `--use-paged-cache` / `--no-paged-cache` always override the default. • Reasoning mode now defaults ON for every reasoning-capable family; `max_thinking_tokens` is honored and gated on both streaming and non-streaming paths. Full history on Cyber Toolchain: https://cybertoolchain.io/tools/vmlx