SAVRN Model Hub
Open-Weight Models
An open-weight model is an AI model whose trained weights are published for anyone to download. The weights are what the model learned in training. With a copy of them you can run the model on hardware you control and train it further on your own data.
Open weights are not the same as open source. Many publishers release the weights without the training data or code, and the license sets what you may do with the model. This library puts each model's full card, architecture, files, license and published evaluations on one page.
Updated 2026-10-19 · How the library is built
11,442 models, sorted by most downloaded.
This repository is under construction. No selected, validated model has been released yet. This model was created by Thea Tollersrud and Per Egil Kummervold in the NB-ASR project. The planned model recognizes equivalent Bokmål–Nynorsk text while detecting changed numbers, pronouns, entities, and participant roles. Meaning-preserving differences and reorderings are accepted. Training starts from Qwen/Qwen3-Reranker-0.6B and uses revision 619e36b85e4943f23cd85b66edcde331995e4872. Validation is automatic, without manual review. Accuracy will be conditional on the supplied positive pairs and the automatic corruption rules. See trainingstatus.json for the latest recorded job status. A complete…
Public original checkpoint shards with full persistent model/optimizer state and publication metadata. Includes existing ~10B-input-token milestones and finals; this is not a claim that cancelled or ongoing runs finished. Source inventories pin regions, object generations, byte counts and available receipt SHA256 hashes. Per-object upload receipts distinguish verified delivery from planned scope. Historical data and scientific settings are not rewritten.
Public original checkpoint shards with full persistent model/optimizer state and publication metadata. Includes existing ~10B-input-token milestones and finals; this is not a claim that cancelled or ongoing runs finished. Source inventories pin regions, object generations, byte counts and available receipt SHA256 hashes. Per-object upload receipts distinguish verified delivery from planned scope. Historical data and scientific settings are not rewritten.
Public original checkpoint shards with full persistent model/optimizer state and publication metadata. Includes existing ~10B-input-token milestones and finals; this is not a claim that cancelled or ongoing runs finished. Source inventories pin regions, object generations, byte counts and available receipt SHA256 hashes. Per-object upload receipts distinguish verified delivery from planned scope. Historical data and scientific settings are not rewritten.
Released tokenizer and generator pairs for Next-Concept Modeling. Each folder holds a config.yaml with the architecture and sampling settings and the.pt checkpoints it names.
Nebium-Small is a 117-million-parameter causal Transformer trained for autoregressive next-chess-move prediction over Lichess UCI move sequences. - Rotary Position Embeddings (RoPE) on attention query and key projections ($\theta = 10000$) - SwiGLU feed-forward transformation - RMSNorm pre-normalization - Causal mask with padding token masking - Byte-Pair Encoding (BPE) tokenizer trained on UCI move plies $$L(N, D) = 1.69 + \frac{406.4}{N^{0.34}} + \frac{410.7}{D^{0.28}}$$ MIT License.
Quantized and FP16 GGUF format binaries for Nebium-Small (117M parameters). Designed for low-latency CPU and GPU execution with llama.cpp and Ollama. MIT License.
Glade is a native Swift library for on-device speech inference on Apple silicon. This repository supplies converted model assets and configuration for Glade's model-specific host runtime. Glade's runtime source is currently private. FP16 Glade conversion of NVIDIA Nemotron 3 Diarization, source revision f667ed73aee57d40cc39428eb768b4fd87a0a29e, under OpenMDW-1.1. See LICENSE and NOTICE. This conversion is not endorsed by NVIDIA. Two self-contained bundles preserve streaming speaker identities and support An encoder position represents 80 ms of audio. Both policies update their speaker cache every 40 positions. Initial buffering is approximately 2.88 / 10.56 seconds. These are explicit…
This is an OpenVINO IR conversion of NVIDIA's nvidia/Nemotron-3-Diarization speaker diarization model ("who spoke when"). It was made for Scrybe, a local Windows transcription app, to add speaker labels ("Speaker 1", "Speaker 2",...) to transcripts on Intel CPUs and GPUs. Not affiliated with or endorsed by NVIDIA. This is an independent conversion. For the original model, its training data, evaluation and responsible-use information, see NVIDIA's model card. The IR holds one offline processing step of the model: the feature embedder, the 31-layer encoder, the projection, the upsampler and the speaker classifier. It does not include the audio front end or the speaker-cache logic. The calling…
Glade is a native Swift library for on-device speech inference on Apple silicon. This repository supplies converted model assets and configuration for Glade's model-specific host runtime. Glade's runtime source is currently private. Glade assets using Apple Core AI APIs, derived from nvidia/nemotron-3.5-asr-streaming-0.6b, revision ea30d66debe3740a08b573244286791d423d6b3e. Targets physical Apple silicon devices running macOS 27 or iOS 27. These are source.aimodel assets that specialize on the device, not Core ML models or precompiled device-specific binaries. The bundle contains an ANE-oriented W8A16 streaming encoder, FP16 causal subsampling, FP16 ANE-oriented recurrent predictor and…
A browser-friendly build of nvidia/nemotron-3.5-asr-streaming-0.6b, NVIDIA's multilingual cache-aware FastConformer RNNT with native streaming, punctuation, capitalization, and language-ID prompting (40 language-locales, 19 transcription-ready). Used by Twitch Live Captions for on-device live captions. It is derived from the sherpa-onnx export csukuangfj2/sherpa-onnx-nemotron-3.5-asr-streaming-0.6b-560ms-2026-06-11. The encoder's MatMul weights are quantized to 4 bits with ONNX Runtime's MatMulNBitsQuantizer (block size 32, symmetric), shrinking it from 2.5 GB to 683 MB. The decoder and joiner are unchanged fp32. Auto-detection can misfire at the start of an utterance (for example, a few…
Glade is a native Swift library for on-device speech inference on Apple silicon. This repository supplies converted model assets and configuration for Glade's model-specific host runtime. Glade's runtime source is currently private. Glade assets using Apple Core AI APIs, derived from nvidia/nemotron-speech-streaming-en-0.6b, revision ebe59e5a817142986528bbbee5dba8db7b38ed50. Targets physical Apple silicon devices running macOS 27 or iOS 27. These are source.aimodel assets that specialize on the device. The bundle shares weights across 1.12-second quality and 160-millisecond preview entrypoints. It contains a W8A16 streaming encoder, FP16 causal subsampling, FP16 recurrent predictor and…
A browser-friendly build of nvidia/nemotron-speech-streaming-en-0.6b, a cache-aware FastConformer RNNT model with native streaming, punctuation, and capitalization. Used by Twitch Live Captions for on-device live captions. It is derived from the sherpa-onnx export csukuangfj2/sherpa-onnx-nemotron-speech-streaming-en-0.6b-560ms-2026-04-25. The encoder's MatMul weights are quantized to 4 bits with ONNX Runtime's MatMulNBitsQuantizer (block size 32, symmetric), shrinking it from 2.4 GB to 680 MB. The decoder and joiner are unchanged fp32. Earlier revisions of this repo (before this card mentioned 560 ms) contain the 1.12 s chunk build. Licensed by NVIDIA Corporation under the NVIDIA Open Model…
Distribución de Neo TTS Studio 1.2 para Linux y Windows, en CPU y CUDA. Los cuatro paquetes son archivos individuales completos, sin partes. Incluyen motor de entrenamiento e inferencia C++17, exportación y panel Axolotl. No contienen modelos de voz privados, credenciales ni datasets. Comprueba las sumas de SHA256SUMS.txt después de descargar y antes de ejecutar. La edición CUDA necesita GPU NVIDIA y controladores compatibles; CUDA Windows está compilada con LibTorch CUDA 12.8, pero aún no se ha probado en un equipo Windows con NVIDIA. En Windows abre PowerShell en la carpeta del ZIP extraído:./neo-tts.exe --webui --port 8787. En Linux ejecuta./run.sh --webui --port 8787 --data…
Reading notes and an experiment sketch for Neural Architecture Search. The repository emphasizes what still needs to be tested instead of manufacturing scores or release claims. - the scope of the research question and likely confounders - a proposed comparison with matched baselines - concrete evaluation context such as task-appropriate public benchmarks named in the main note - reproducibility checks, failure modes, and open questions - topic-relevant references Start with reading.md for the full note. Sections labeled as plans or hypotheses should not be interpreted as experimental results. If results are added later, they should include dataset versions, commands, seeds, hardware, and…
Model Collections
Hand-picked starting points, each with the reason it exists.
Collection · 4 entries
Embedding models for retrieval
Sentence and document embedding models used to build retrieval systems. Dimension and sequence length matter more than size here, and both come from the publisher.
Collection · 4 entries
Models that fit on one accelerator
Models whose publisher-reported parameter count puts them within reach of a single accelerator at common precisions. Memory needed depends on precision and serving configuration, so treat the parameter count as the starting point, not the answer.
Collection · 6 entries
Open-weight text models worth knowing
Widely used open-weight language models, chosen because each one is a distinct family rather than a variant of the one above it. Selection, not a ranking.
Collection · 3 entries
Speech and audio models
Recognition and synthesis models, grouped so the two directions are easy to compare.
Open-Weight Models Explained
What is an open-weight model?
An AI model whose trained weights are published for anyone to download, so it can be run, tested and fine-tuned on hardware the user controls.
Is an open-weight model the same as open source?
Not always. Open weights means the trained model can be downloaded. Open source usually also means the training code and data are available and the license allows broad reuse. Many open-weight models release the weights only.
Can I use an open-weight model commercially?
It depends on the license. Apache 2.0 and MIT allow commercial use. Other licenses limit it, for example to non-commercial use or below a set number of users. Every model page here shows its license.
How much memory does an open-weight model need?
About two bytes per parameter at 16-bit precision, so a 7-billion-parameter model needs roughly 14 GB for its weights, plus memory for the context it processes. Each model page lists its parameter count and the size of its files.
Related SAVRN Research
The hub sits beside SAVRN's market data and infrastructure research: what models cost to run, and what it takes to run them.
SAVRN Index
What open models cost to run
The same open-weight model priced by every host that serves it, per million tokens.
Research Hub
Data center trackers and maps
Moratoriums, permits, power, water and capital behind the facilities that run these models.
Method
How the Model Hub is built
Sources, evidence labels, refresh behaviour, and the limits of every comparison here.