Repository navigation
Releases
Releases · MIT-RLX/rlx-models
v0.2.19
The first release since 0.2.11. The 0.2.14 and 0.2.16 releases were prepared but never published, so their changes are included here. Every crate in the workspace is versioned 0.2.19 and built against upstream rlx 0.2.19.
Full notes: CHANGELOG.md § 0.2.19
Upstream rlx 0.2.19
- Every
rlx*pin is^0.2.19, includingrlx-programin the standaloneprogram/manifests. - All 32 pinned upstream crates resolve from crates.io (including
rlx-torch-ckptandrlx-opscope). A fresh clone builds without a sibling../rlxcheckout. scripts/publish.shtier fixes:rlx-sanottsmoved to tier 2, andrlx-models-pluginnow has its own final tier after the facade.
New model families (highlights)
- LLM / MoE: Recursive Flint (40B-A2.5B, experts paged from disk), DeepSeek-V4.1-Flash, Qwen3.8-Flash-Next / Qwen4 arch (
rlx-qwen4exp), GLM-5.3-Flash (rlx-glm5next), Kimi-K3 (training, inspection, distributed), Ling 3.0, Motif-3, Jamba, Llama 4, Llama-3.2-Vision, MiniCPM-V / MiniCPM-SALA, Nanbeige4.2, TinyLlama, Laguna, Ternary-Bonsai-2-27B, Perplexity pplx-decider - TTS / audio: Fish S2-Pro + Drama 3, KittenTTS 2, Pocket TTS, Kokoro, Parler-TTS, Soprano, MOSS-TTS-Nano, TADA, sanoTTS, ChatterBox, Orpheus, and many more. TTS crates are now rlx-native (ort-free).
- ASR / speech: Moonshine v2 streaming, Conformer-CTC, Parakeet-TDT, VibeVoice-ASR, Qwen3-ASR, TEN-VAD (bit-identical frontend, embedded/MCU targets), wake-word suite
- Vision / image: FLUX.1, FLUX 3 Klein, Open Genmoji, DINOv3, SigLIP 2, UNI2-h, BiRefNet,
rlx-upscale(15 SR architectures), TRELLIS.2, Unlimited-OCR / Jina-OCR, PP-OCRv6 - Embeddings: EmbeddingGemma-300M, jina-embeddings-v5-text-nano
- Time series: Mistral Timestral, TimesFM 3
Platform
- One exporter for every family:
rlx_core::export/rlx-export(.rlxppacks) - Number formats:
--number-format f16|bf16|f8|f6|f4|fNeXmY|f64|f64x2on every LM CLI, NVFP4-style blocks, f64 double-single on GPU backends. QAT and fine-tuning viarlx-run qat. - Serving:
rlx-openaimulti-model OpenAI HTTP server and therlx-servelibrary. Host-driven decode now covers 8 families (up from 3), andlm_registrylets any crate register a model withauto_runner. - Browser:
rlx-models-web(wasm + WebGPU). Qwen3.5 runs split by layer across Web Workers. - Decoding: decoders share
rlx-search. Also one single-bucket decode cache, and stop-signal fixes in five families. - Memory: streamed packed-weight uploads (qwen3/qwen35/gemma/lfm/dflash: RSS reduced by 4.7–19.6 GB) and a lazy
token_embd. - Backends: many correctness fixes across CPU, Metal, MLX, CUDA, ROCm, wgpu and Vulkan. Experimental
tpuandcoremldevice features. - Tooling:
just lint-features(non-default feature states) andjust release-preflight.
Known issues
- CPU arena reuse miscompiles a real-scale DeepSeek-V4.1 attention graph (see CHANGELOG).
v0.2.9
First public release of the rlx-models model crates on crates.io.
All model crates are published at 0.2.9, aligned with the rlx 0.2.9 compiler + runtime framework. They run natively across the rlx multi-backend runtime (CPU / Metal / MLX / CUDA / ROCm / wgpu / Vulkan / CoreML).
Scope — 80+ model ports across:
- ASR — Whisper, FunASR, NeMo/Nemotron, Qwen3-ASR, Voxtral, wav2vec2
- TTS — Kitten, Orpheus, NeuTTS, Inflect-Nano, Kyutai, Qwen3-TTS, VibeVoice
- Audio codecs — Mimi, DAC, EnCodec, SNAC, TSAC, SpeechTokenizer, WavTokenizer, XCodec
- LLM / VLM — Gemma, Qwen3 / Qwen3-VL, Llama 3.2, Phi, GLM, Mistral, Cohere, GPT-OSS, Mamba, Eagle3, and more
- Vision — SAM/SAM2/SAM3, DINOv2, Grounding-DINO, Florence-2, BioCLIP2
Engineering
- Workspace-wide
cargo fmt+cargo clippy --all-targets -D warningsclean. - Embedded build assets shipped in-crate; published in dependency (tier) order via
scripts/publish.sh.