Gemma 4 E4B-IT uncensored by HauhauCS. 0/465 Refusals\ No changes to datasets or capabilities. Fully functional, 100% of what the original authors intended - just without the refusals. These are meant to be the best lossless uncensored models out there. Stronger uncensoring — model is fully unlocked and won't refuse prompts. May occasionally append short disclaimers (baked into base model training, not refusals) but full content is always generated. For a more conservative uncensor that keeps some safety guardrails, check the Balanced variant when it's available. All quants generated with importance matrix (imatrix) for optimal quality preservation on abliterated weights. KP ("Perfect")…
SAVRN Model Hub · Models by Task
Image and Text to Text Models
138 open-weight image and text to text models in the SAVRN Model Hub, with Qwen, Unsloth AI and OpenVLA Collaboration publishing the most.
138Models
71Publishers
3M to 2.8TParameter range
8Licenses
SAVRN's Take
Hand one of these models a photograph or a screenshot, ask a question, and it answers in text. Our hub lists 138 models for the job, 49 of them in the Index. Qwen publishes 28 and Unsloth AI 11, with Google, LM Studio Community and the OpenVLA Collaboration at 4 each. Downloads follow suit: Qwen3-VL-8B-Instruct pulled 19,098,599 last month, ahead of Qwen3.6-35B-A3B-FP8 at 10,009,539 and Google's gemma-4-26B-A4B-it at 9,794,984.
All eight leaders share a 262,144-token context and fit on a single MI300X at $1.85 an hour on the Index, so the question is how much of that card each takes. At 16-bit there are two tiers: Qwen3.5-4B, Qwen3-VL-8B-Instruct and Qwen3.5-9B need 11.2 to 23.2 GB, leaving the rest of the card for batch, while gemma-4-26B-A4B-it, both Qwen3.8-27B builds, gemma-4-31B-it and Qwen3.6-35B-A3B-FP8 need 61.9 to 86.3 GB. Halve those for 8-bit; 4-bit puts the whole group under 22 GB. Kimi-K3, at 2.8T parameters, is the outlier: 6,671.8 GB at 16-bit, which no Index host can hold, and 1,668.0 GB at 4-bit across 7 MI325X cards at $14.00 an hour.
Licensing is the easy part. 89 of the 138 are Apache 2.0, including all eight leaders, and 16 more are MIT, so most of the catalog deploys without a negotiation. The 20 marked other, Kimi-K3 among them, and the 5 under Gemma terms need a read first; 3 state no license, which we treat as a no. Settle the memory tier first, then the license, then treat the download count as a tally of operators who got there before you.
Most Downloaded
| Model | Publisher | Parameters | License | Monthly downloads | Cheapest GPUs at 16-bit |
|---|---|---|---|---|---|
| Qwen3-VL-8B-Instruct | Qwen | 8.8B | apache-2.0 | 19.1M | 1x MI300X, $1.85/hr |
| Qwen3.6-35B-A3B-FP8 | Qwen | 36B | apache-2.0 | 10M | 1x MI300X, $1.85/hr |
| gemma-4-26B-A4B-it | 25.8B | apache-2.0 | 9.8M | 1x MI300X, $1.85/hr | |
| Qwen3.5-9B | Qwen | 9.7B | apache-2.0 | 9.3M | 1x MI300X, $1.85/hr |
| gemma-4-31B-it | 31.3B | apache-2.0 | 9M | 1x MI300X, $1.85/hr | |
| Qwen3.8-27B | Qwen | 27.8B | apache-2.0 | 7.4M | 1x MI300X, $1.85/hr |
| Qwen3.8-27B-FP8 | Qwen | 27.8B | apache-2.0 | 7.3M | 1x MI300X, $1.85/hr |
| Qwen3.5-4B | Qwen | 4.7B | apache-2.0 | 7M | 1x MI300X, $1.85/hr |
| Qwen2.5-VL-7B-Instruct | Qwen | 8.3B | apache-2.0 | 6.9M | 1x MI300X, $1.85/hr |
| Qwen3.6-27B-FP8 | Qwen | 27.8B | apache-2.0 | 6M | 1x MI300X, $1.85/hr |
Licenses
| License | Models | Commercial use |
|---|---|---|
| apache-2.0 | 89 | Yes |
| other | 20 | Read the license |
| mit | 16 | Yes |
| gemma | 5 | Yes, with conditions |
| openrail | 3 | Yes, with conditions |
| not stated | 3 | Not stated |
| llama2 | 1 | Read the license |
| cc-by-4.0 | 1 | Yes |
Who Publishes Them
| Publisher | Models |
|---|---|
| Qwen | 28 |
| Unsloth AI | 11 |
| OpenVLA Collaboration | 4 |
| LM Studio Community | 4 |
| 4 | |
| Hao Liang | 3 |
All 138 Models, Page 2 of 3
In the past five months since Qwen2-VL’s release, numerous developers have built new models on the Qwen2-VL vision-language models, providing us with valuable feedback. During this period, we focused on building more useful vision-language models. Today, we are excited to introduce the latest addition to the Qwen family: Qwen2.5-VL. Understanding long videos and capturing events: Qwen2.5-VL can comprehend videos of over 1 hour, and this time it has a new ability of cpaturing event by pinpointing the relevant video segments. Capable of visual localization in different formats: Qwen2.5-VL can accurately localize objects in an image by generating bounding boxes or points, and it can provide…
Gemma is a family of open models built by Google DeepMind. Gemma 4 models are multimodal, handling text and image input (with audio supported on small models) and generating text output. This release includes open-weights models in both pre-trained and instruction-tuned variants. Gemma 4 features a context window of up to 256K tokens and maintains multilingual support in over 140 languages. Featuring both Dense and Mixture-of-Experts (MoE) architectures, Gemma 4 is well-suited for tasks like text generation, coding, and reasoning. The models are available in four distinct sizes: E2B, E4B, 26B A4B, and 31B. Their diverse sizes make them deployable in environments ranging from high-end phones…
We're excited to unveil Qwen2-VL, the latest iteration of our Qwen-VL model, representing nearly a year of innovation. SoTA understanding of images of various resolution & ratio: Qwen2-VL achieves state-of-the-art performance on visual understanding benchmarks, including MathVista, DocVQA, RealWorldQA, MTVQA, etc. Understanding videos of 20min+: Qwen2-VL can understand videos over 20 minutes for high-quality video-based question answering, dialog, content creation, etc. Agent that can operate your mobiles, robots, etc.: with the abilities of complex reasoning and decision making, Qwen2-VL can be integrated with devices like mobile phones, robots, etc., for automatic operation based on…
Over recent months, we have intensified our focus on developing foundation models that deliver exceptional utility and performance. Qwen3.5 represents a significant leap forward, integrating breakthroughs in multimodal learning, architectural efficiency, reinforcement learning scale, and global accessibility to empower developers and enterprises with unprecedented capability and efficiency. For more details, please refer to our blog post Qwen3.5. WMT24++: a harder subset of WMT24 after difficulty labeling and rebalancing; we report the averaged scores on 55 languages using XCOMET-XXL. Empty cells (--) indicate scores not yet available or not applicable. Empty cells (--) indicate scores not…
Following the widespread community adoption of the Qwen3.5 and Qwen3.6 series, we are pleased to introduce Qwen3.8, the most capable generation in the Qwen open-model family to date. Built on the architectural foundation of Qwen3.5, Qwen3.8 delivers substantial gains across coding, professional work, research, and long-horizon agentic tasks. Qwen3.8-27B brings these advances to a compact, deployment-friendly dense model: a native vision-language model that understands images and videos, with flexible thinking control, designed to carry complex, multi-step tasks through to completion with greater reliability. For streamlined integration, we recommend using Qwen3.8 via APIs. Qwen3.8 can be…
You can now also fine-tune the model locally with Unsloth. - Read our Qwen3.5 fine-tuning guide here. Over recent months, we have intensified our focus on developing foundation models that deliver exceptional utility and performance. Qwen3.5 represents a significant leap forward, integrating breakthroughs in multimodal learning, architectural efficiency, reinforcement learning scale, and global accessibility to empower developers and enterprises with unprecedented capability and efficiency. For more details, please refer to our blog post Qwen3.5. WMT24++: a harder subset of WMT24 after difficulty labeling and rebalancing; we report the averaged scores on 55 languages using XCOMET-XXL. Empty…
As the frontier of foundation models pushes toward ever-larger parameter counts and ever-longer context windows, the question is no longer just how much we can scale, but how efficiently we can do so. Sustainable progress toward artificial general intelligence (AGI) that benefits everyone demands architectural innovation. Today, we are sharing a concrete step in that direction: Qwen3.8-Flash-Next. This experimental preview of the architecture that will underpin Qwen4 is built around a fundamental rethinking of how the core components of modern large language models (LLMs) interact at scale. The first open-weight release under this architecture is Qwen3.8-Flash-Next, which introduces: For…
Model · Image and text to text
Qwen3.5-9B-The-Defiant-Fable-Uncensored-Heretic-NEO-IMATRIX-MAX-MTP-GGUF
and it does so in 4bit and 8bit. Regular and MTP (fast) NEO IMATRIX GGUFs provided. (this model is part of the Qwen 3.6 27B Fable Fusion 711 pipelines: 2200+ likes, 3 million + downloads) instruct modes (2 new - Spoon / Einstein, all use ZERO REASONING TOKENS) all switchable on the fly via API, direct and "in chat" (yes - model ctrl at the chat/message level). Model name has "plusIQ" in the name. (there is also a extra robust "tools" version too.) Extreme intelligence in a small package. Jaw dropping performance. Superior instruction following. A multi-stage and multi-model fine tune and multi-stage merge on local hardware by myself and Nightmedia. Several of my 9B Qwen 3.5 fine tunes were…
Inkling-Small is a general-purpose multimodal model that accepts text, image and audio inputs and generates text outputs. It is intended for use in English and other languages, and across multiple coding languages. The model is designed to be used by developers building AI-powered applications, including agentic and tool-use systems, coding assistants, chatbots, and retrieval-augmented generation systems, and is suitable for general-purpose conversational use, instruction-following, and other natural language and multimodal tasks. It is released with open weights to support research, fine-tuning and integration into third-party products by downstream developers. Languages: English, with…
AWQ 4-bit (W4A16) quantization of baidu/Unlimited-OCR, a 3B vision-language OCR model that pushes DeepSeek-OCR one step further (one-shot, long-horizon document parsing). This repo quantizes the DeepSeek-V2 MoE text decoder with activation-aware scaling (AWQ) while keeping the vision tower in BF16, so it stays a drop-in transformers model. Unlimited-OCR uses the DeepSeek-OCR prompt vocabulary. The prompt must contain; prefix it with whenever you also want bounding boxes for what was read. - base — basesize=1024, imagesize=1024, cropmode=False. Good default for normal pages. - gundam — basesize=1024, imagesize=640, cropmode=True. Tiles the page; use for dense or large/high-resolution…
Run with https://llama.app - https://huggingface.co/Qwen/Qwen3.8-27B - add info
Following the February release of the Qwen3.5 series, we're pleased to share the first open-weight variant of Qwen3.6. Built on direct feedback from the community, Qwen3.6 prioritizes stability and real-world utility, offering developers a more intuitive, responsive, and genuinely productive coding experience. This release delivers substantial upgrades, particularly in For more details, please refer to our blog post Qwen3.6-35B-A3B. Empty cells (--) indicate scores not available or not applicable. For streamlined integration, we recommend using Qwen3.6 via APIs. Below is a guide to use Qwen3.6 via OpenAI-compatible API. Qwen3.6 can be served via APIs with popular inference frameworks. In…
This is the QAT INT4 Flax checkpoint (from Kaggle) converted to HF+AWQ format for ease of use. AWQ was NOT used for quantization. You can find the conversion script convertflax.py in this model repo. NOTE: this is NOT the same as the official QAT INT4 GGUFs released here https://huggingface.co/collections/google/gemma-3-qat-67ee61ccacbf2be4195c265b Below is the original Model card from https://huggingface.co/google/gemma-3-27b-it [Gemma 3 Technical Report][g3-tech-report] [Responsible Generative AI Toolkit][rai-toolkit] [Gemma on Kaggle][kaggle-gemma] [Gemma on Vertex Model Garden][vertex-mg-gemma3] Summary description and brief definition of inputs and outputs. Gemma is a family of…
Set -DGGMLCUDA=OFF for CPU/Metal. -np > 1 and --mmproj are not yet supported with MTP. - Developer Role Support so Qwen3.6 can work in Codex, OpenCode and more! - Qwen3.6 can now be run and fine-tuned in Unsloth Studio. Read our guide. Following the February release of the Qwen3.5 series, we're pleased to share the first open-weight variant of Qwen3.6. Built on direct feedback from the community, Qwen3.6 prioritizes stability and real-world utility, offering developers a more intuitive, responsive, and genuinely productive coding experience. This release delivers substantial upgrades, particularly in For more details, please refer to our blog post Qwen3.6-27B. For streamlined integration…
Model · Image and text to text
Qwen3.8-27B-TURBO-Fable-Cold-Fusion-735-882-Heretic-Uncensored-NEO-CODER-MAX-MTP-GGUF
in 8 bit and over 718 arc-c in 4 bit. This version is called TURBO because it drastically reduces thinking tokens (by 1/2 to as high as 1/10), yet maintains output detail and quality. In otherwords while "reg" Qwen3.8 27B is thinking about "formatting" for a few 1000 tokens, this model is already done and waiting for more. This repo contains both "regular" and "MTP" Neo-CODER MAX DI-MATRIX (duel imatrix) GGUF quants. NOTE: Please see the "community" tab for user experiences, additional third party benchmarks (including strongest tool calling performance ever recorded), and other quant versions (also see "Quantized" in the right "model tree" too). The strongest, smartest open source…
In addition to the original formula, we have further enhanced Qwen2.5-VL-32B's mathematical and problem-solving abilities through reinforcement learning. This has also significantly improved the model's subjective user experience, with response styles adjusted to better align with human preferences. Particularly for objective queries such as mathematics, logical reasoning, and knowledge-based Q&A, the level of detail in responses and the clarity of formatting have been noticeably enhanced. In the past five months since Qwen2-VL’s release, numerous developers have built new models on the Qwen2-VL vision-language models, providing us with valuable feedback. During this period, we focused on…
Surya is a 650M param OCR model with these features: - Accuracy - scores 83.3% on olmOCR-bench (top under 3B params) - Multilingual - scores 87.2% on an internal benchmark set of 91 languages (more here) - Layout analysis (table, image, header, etc.) with reading order - Table recognition (rows + columns) It works on a range of documents (see usage and benchmarks). Our managed platform runs both Surya, and variants of our highest accuracy model, Chandra. Get started with $5 in free credits — sign up (takes under 30 seconds) or try our free public playground. Surya is named for the Hindu sun god, who has universal vision. The Surya code is licensed under Apache 2.0. The model weights use a…
Following the February release of the Qwen3.5 series, we're pleased to share the first open-weight variant of Qwen3.6. Built on direct feedback from the community, Qwen3.6 prioritizes stability and real-world utility, offering developers a more intuitive, responsive, and genuinely productive coding experience. This release delivers substantial upgrades, particularly in For more details, please refer to our blog post Qwen3.6-27B. For streamlined integration, we recommend using Qwen3.6 via APIs. Below is a guide to use Qwen3.6 via OpenAI-compatible API. Qwen3.6 can be served via APIs with popular inference frameworks. In the following, we show example commands to launch OpenAI-Compatible API…
Qwen3.6-35B-A3B uncensored by HauhauCS. 0/465 Refusals. No changes to datasets or capabilities. Fully functional, 100% of what the original authors intended - just without the refusals. These are meant to be the best lossless uncensored models out there. Stronger uncensoring — model is fully unlocked and won't refuse prompts. May occasionally append short disclaimers (baked into base model training, not refusals) but full content is always generated. For a more conservative uncensor that keeps some safety guardrails, check the Balanced variant when it's available. All quants generated with importance matrix (imatrix) for optimal quality preservation on abliterated weights. KP ("Perfect")…
Over recent months, we have intensified our focus on developing foundation models that deliver exceptional utility and performance. Qwen3.5 represents a significant leap forward, integrating breakthroughs in multimodal learning, architectural efficiency, reinforcement learning scale, and global accessibility to empower developers and enterprises with unprecedented capability and efficiency. For more details, please refer to our blog post Qwen3.5. WMT24++: a harder subset of WMT24 after difficulty labeling and rebalancing; we report the averaged scores on 55 languages using XCOMET-XXL. Empty cells (--) indicate scores not yet available or not applicable. Empty cells (--) indicate scores not…
This repo quantizes the model using data-free quantization technique. As of 2026-02-25, make sure your system has cuda12.8 installed. Then, create a fresh Python environment (e.g. python3.12 venv) and run: Over recent months, we have intensified our focus on developing foundation models that deliver exceptional utility and performance. Qwen3.5 represents a significant leap forward, integrating breakthroughs in multimodal learning, architectural efficiency, reinforcement learning scale, and global accessibility to empower developers and enterprises with unprecedented capability and efficiency. For more details, please refer to our blog post Qwen3.5. WMT24++: a harder subset of WMT24 after…
Model · Image and text to text
Qwen3.8-27B-GSQ-RCO-GGUF
Non-uniform GGUF quantizations produced with GSQ and RCO, with a vision projector for multimodal use. This repository provides GGUF quantizations of Qwen3.8-27B at four sizes, together with the model's vision projector (mmproj) for multimodal use. In contrast to uniform quantization, which applies a single quantization type to all weight tensors, each model here assigns a separate quantization type to every tensor. The assignment is obtained by a gradient-based search that allocates precision according to per-tensor sensitivity, subject to a total size budget. The resulting files are standard GGUF and run unmodified in llama.cpp, Ollama, and LM Studio. Both methods were developed at the…
Using llama.cpp release b10262 for quantization. Don't know which to choose? Grab Q4KM (21.86GB) - usually a good mix of size and performance. Download instructions available here First, make sure you have the Hugging Face CLI installed: The files marked true in the Split column above are stored as multiple parts in a folder. To download all the parts to a local folder, run: You can either specify a new local-dir (endless-frontierBigBang-v1-bf16) or download them all in place (./) These quants run with llama.cpp - installable in one line via llama.app: llama-server includes a built-in chat web UI, served at http://localhost:8080 by default. These quants were made with llama.cpp release…
Set -DGGMLCUDA=OFF for CPU/Metal. -np > 1 and --mmproj are not yet supported with MTP. - Developer Role Support so Qwen3.6 can work in Codex, OpenCode and more! - Qwen3.6 can now be run and fine-tuned in Unsloth Studio. Read our guide. Following the February release of the Qwen3.5 series, we're pleased to share the first open-weight variant of Qwen3.6. Built on direct feedback from the community, Qwen3.6 prioritizes stability and real-world utility, offering developers a more intuitive, responsive, and genuinely productive coding experience. This release delivers substantial upgrades, particularly in For more details, please refer to our blog post Qwen3.6-35B-A3B. Empty cells (--) indicate…
FP8-dynamic variant of gemma-4-26B-A4B-it.
You can now also fine-tune the model locally with Unsloth. - Read our Qwen3.5 fine-tuning guide here. Over recent months, we have intensified our focus on developing foundation models that deliver exceptional utility and performance. Qwen3.5 represents a significant leap forward, integrating breakthroughs in multimodal learning, architectural efficiency, reinforcement learning scale, and global accessibility to empower developers and enterprises with unprecedented capability and efficiency. For more details, please refer to our blog post Qwen3.5. WMT24++: a harder subset of WMT24 after difficulty labeling and rebalancing; we report the averaged scores on 55 languages using XCOMET-XXL. Empty…
Following the February release of the Qwen3.5 series, we're pleased to share the first open-weight variant of Qwen3.6. Built on direct feedback from the community, Qwen3.6 prioritizes stability and real-world utility, offering developers a more intuitive, responsive, and genuinely productive coding experience. This release delivers substantial upgrades, particularly in For more details, please refer to our blog post Qwen3.6-27B. Empty cells (--) indicate scores not yet available or not applicable. For streamlined integration, we recommend using Qwen3.6 via APIs. Below is a guide to use Qwen3.6 via OpenAI-compatible API. Qwen3.6 can be served via APIs with popular inference frameworks. In…
Using llama.cpp release b10142 for quantization. All quants made using imatrix option with dataset from here Run them in your choice of tools: Note: if it's a newly supported model, you may need to wait for an update from the developers. Some of these quants (Q3KXL, Q4KL etc) are the standard quantization method with the embeddings and output weights quantized to Q80 instead of what they would normally default to. First, make sure you have huggingface-cli installed: Then, you can target the specific file you want: If the model is bigger than 50GB, it will have been split into multiple files. In order to download them all to a local folder, run: You can either specify a new local-dir…
Model · Image and text to text
Qwen3.6-27B-Fable-Fusion-711-Uncensored-Heretic-NM-DAU-NEO-MAX-MTP-GGUF
in both 8 bit and 4 bit. This repo contains both "regular" and "MTP" Neo MAX Imatrix GGUF quants. Many other additional quant types avail too. 3rd parties confirm this model's performance in the "community tab". 40B versions: Eleanor-DECKARD and Grand Intelligence - FF711-717 || Qwen 3.8 27B Cold Fusion (1/2 to 1/10 thinking size, more brainpower): COLD FUSION Meet the newest, strongest and fastest Qwen 3.8: The TURBO Fable 738-882 The strongest, smartest open source multi-stage model fine tune for consumer hardware ever and BUILT on consumer hardware via Unsloth. The first model of this size/type to breach "700" ARC-C in both 8 bit and 4 bit; hench the "711" in the name. This model (both 4…
This is an uncensored version of Qwen/Qwen3.6-27B created with abliteration (see remove-refusals-with-transformers to know more about it). This is a crude, proof-of-concept implementation to remove refusals from an LLM model without using TransformerLens. - AWQ Marlin kernel supported (auto-converted by vLLM at runtime) - MTP speculative decoding supported out of the box - 110+ tok/s on a single A800 80GB (vLLM 0.21.0, MTP enabled, fp8 KV cache) - Risk of Sensitive or Controversial Outputs: This model’s safety filtering has been significantly reduced, potentially generating sensitive, controversial, or inappropriate content. Users should exercise caution and rigorously review generated…
Over recent months, we have intensified our focus on developing foundation models that deliver exceptional utility and performance. Qwen3.5 represents a significant leap forward, integrating breakthroughs in multimodal learning, architectural efficiency, reinforcement learning scale, and global accessibility to empower developers and enterprises with unprecedented capability and efficiency. For more details, please refer to our blog post Qwen3.5. WMT24++: a harder subset of WMT24 after difficulty labeling and rebalancing; we report the averaged scores on 55 languages using XCOMET-XXL. Empty cells (--) indicate scores not yet available or not applicable. MathVision：our model’s score is…
Gemma is a family of open models built by Google DeepMind. Gemma 4 models are multimodal, handling text and image input (with audio supported on E2B, E4B, and 12B) and generating text output. This release includes open-weights models in both pre-trained and instruction-tuned variants. Gemma 4 features a context window of up to 256K tokens and maintains multilingual support in over 140 languages. Featuring both Dense and Mixture-of-Experts (MoE) architectures, Gemma 4 is well-suited for tasks like text generation, coding, and reasoning. The models are available in five distinct sizes: E2B, E4B, 12B, 26B A4B, and 31B. Their diverse sizes make them deployable in environments ranging from…
Model · Image and text to text
gemma-3-27b-it-GPTQ-4b-128g
This model was obtained by quantizing the weights of gemma-3-27b-it to INT4 data type. This optimization reduces the number of bits per parameter from 16 to 4, reducing the disk size and GPU memory requirements by approximately 75%. Only the weights of the linear operators within languagemodel transformers blocks are quantized. Vision model and multimodal projection are kept in original precision. Weights are quantized using a symmetric per-group scheme, with group size 128. The GPTQ algorithm is applied for quantization. Model checkpoint is saved in compressedtensors format. This model was evaluated on the OpenLLM v1 benchmarks. Model outputs were generated with the vLLM engine. The…
Tiel is the fast coder of the arsenal. At 4-bit quantization and 23 GB it fixes real codebase issues at the rate (and speed, with the right GPU) of Opus 4.6 medium, while holding the best multi-turn conversation of any local model we have measured. It is also cheerfully bad at trivia. Pick it for work. Pick something else for exams. Where it sits against the other local builds Multi-turn conversation Reasoning and knowledge Where it stands. On 25 SWE-bench-Live problems Tiel fixes 12 — the same as Opus 4.6 (medium), four more than Ornith-1.5 itself, three more than Nail, and four more than Sonnet 5 (medium). Among models of its own class it is first; the ones ahead are dense 27Bs and Opus…
VLX-Seek-1.5-10B is the open-source 10B model in the VLX-Seek 1.5 family, designed for fine-grained perception and visual grounding in embodied scenarios. It targets practical settings such as drones, robots, robotic dogs, surveillance cameras, inspection systems, and other edge-side visual intelligence applications where a model must identify what is present, localize the right instance, and avoid grounding objects that are absent. Unlike coordinate-generation-based VLMs that directly decode bounding-box numbers, VLX-Seek reformulates localization as region retrieval and region reference. Candidate visual regions are represented as addressable entities, and the model answers by selecting…
Model · Image and text to text
Qari-OCR-v0.3-VL-2B-Instruct
QARI-OCR v0.3 is a specialized vision-language model fine-tuned for Arabic Optical Character Recognition with a focus on structural document understanding. - Built on Qwen2-VL-2B-Instruct, this model excels at preserving document layouts, HTML tags, and formatting while transcribing Arabic text. - It is described in detail in the paper QARI-OCR: High-Fidelity Arabic Text Recognition through Multimodal Large Language Model Adaptation. While QARI v0.2 achieves better raw text accuracy (CER: 0.061), QARI v0.3 excels in: - HTML/Markdown structure preservation - Document layout understanding - Handwritten text recognition (initial capabilities) - 5x faster training than v0.2 You can load this…
Model · Image and text to text
gemma-4-Ortenzya-The-Creative-Wordsmith-31B-it-uncensored-heretic-GGUF
Creating these models takes significant time, work and compute. If you find them useful consider supporting me: Your help will motivate me and would go into further improving my workflow and coverings fees for storage, compute and may even help uncensoring bigger model with rental Cloud GPUs. GGUF quantizations of llmfan46/gemma-4-Ortenzya-The-Creative-Wordsmith-31B-it-uncensored-heretic. attn.oproj Lower refusals indicate fewer content restrictions, while lower KL divergence indicates more closeness to the original model's baseline. Higher refusals cause more rejections, objections, pushbacks, lecturing, censorship, softening and deflections. The aim of this finetune was to improve this…
https://huggingface.co/microsoft/Florence-2-base-ft with ONNX weights to be compatible with Transformers.js. If you haven't already, you can install the Transformers.js JavaScript library from NPM using: Example: Perform image captioning with onnx-community/Florence-2-base-ft. We also released an online demo, which you can try yourself: https://huggingface.co/spaces/Xenova/florence2-webgpu Note: Having a separate repo for ONNX weights is intended to be a temporary solution until WebML gains more traction. If you would like to make your models web-ready, we recommend converting to ONNX using Optimum and structuring your repo like this one (with ONNX weights located in a subfolder named onnx).
PaddleOCR-VL-1.5: Towards a Multi-Task 0.9B VLM for Robust In-the-Wild Document Parsing PaddleOCR-VL-1.5 is an advanced next-generation model of PaddleOCR-VL, achieving a new state-of-the-art accuracy of 94.5% on OmniDocBench v1.5. To rigorously evaluate robustness against real-world physical distortions—including scanning artifacts, skew, warping, screen photography, and illumination—we propose the Real5-OmniDocBench benchmark. Experimental results demonstrate that this enhanced model attains SOTA performance on the newly curated benchmark. Furthermore, we extend the model’s capabilities by incorporating seal recognition and text spotting tasks, while remaining a 0.9B ultra-compact VLM…
ProcVLM-2B is a procedure-grounded vision-language model for estimating progress rewards from robot manipulation observations. Given a task description and a recent window of video frames, the model reasons about the remaining atomic actions and predicts the current task completion percentage. ProcVLM-2B is designed for research on robot learning, progress reward modeling, embodied evaluation, and procedure-aware video understanding. Typical use cases include: - estimating task completion progress from robot videos; - producing dense progress rewards from sparse demonstrations; - adapting progress prediction to a new environment with one-shot LoRA fine-tuning. This model is not intended to…
This model has had its safety alignment substantially removed via abliteration (orthogonalizing the refusal direction out of the residual stream). It will comply with harmful, unethical, or illegal requests the original Qwen3.8-Flash-Next would refuse. Released strictly for legitimate research — interpretability, AI-safety / refusal-mechanism study, red-teaming, and robustness evaluation. You assume full responsibility for how you use it and everything it generates; add your own safety and moderation layers before any deployment. Use must comply with the Apache 2.0 License inherited from the base model and all applicable law. The authors accept no liability for misuse. - A Blackwell GPU…
This model was produced by fine-tuning the OpenVLA 7B model via LoRA (r=32) on the LIBERO-Spatial dataset from the LIBERO simulation benchmark. We made a few modifications to the training dataset to improve final performance (see the OpenVLA paper for details). Below are the hyperparameters we used for all LIBERO experiments: - No gradient accumulation (i.e. gradaccumulationsteps == 1) - shufflebuffersize == 100000 See the OpenVLA GitHub README for instructions on how to run and evaluate this model in the LIBERO simulator.
Moondream 3.1 is a vision language model with a mixture-of-experts architecture (9B total parameters, 2B active). It delivers state-of-the-art visual reasoning and detection while staying fast and cheap to deploy. Skills include query, detect, point, and caption, all native and all returning structured output. For the full story on what's new — including how we trained it and how it holds up on your own tasks — see the release notes. Photon is Moondream's high-performance inference engine. It runs the model locally on NVIDIA GPUs (Ampere or newer) and Apple Silicon Macs, with the same API as Moondream Cloud. No API key is required to run the base model locally. (You'll only need one to run…
This model is Rex-Omni, a 3B-parameter Multimodal Large Language Model (MLLM) presented in the paper "Detect Anything via Next Point Prediction". It is compatible with the Hugging Face transformers library and is licensed under the IDEA License 1.0. src="https://img.shields.io/badge/RexOmni-Website-BADFDB?style=flat-square&logo=deno&logoColor=violet&color=BADFDB" alt="RexThinker Website" src="https://img.shields.io/badge/RexOmni-Paper-Red%25red?logo=arxiv&logoColor=red&color=yellow" alt="RexThinker Paper on arXiv" src="https://img.shields.io/badge/RexOmni-Weight-orange?logo=huggingface&logoColor=yellow" alt="RexThinker weight on Hugging Face"…
This model was produced by fine-tuning the OpenVLA 7B model via LoRA (r=32) on the LIBERO-Object dataset from the LIBERO simulation benchmark. We made a few modifications to the training dataset to improve final performance (see the OpenVLA paper for details). Below are the hyperparameters we used for all LIBERO experiments: - No gradient accumulation (i.e. gradaccumulationsteps == 1) - shufflebuffersize == 100000 See the OpenVLA GitHub README for instructions on how to run and evaluate this model in the LIBERO simulator.
This model was produced by fine-tuning the OpenVLA 7B model via LoRA (r=32) on the LIBERO-10 (LIBERO-Long) dataset from the LIBERO simulation benchmark. We made a few modifications to the training dataset to improve final performance (see the OpenVLA paper for details). Below are the hyperparameters we used for all LIBERO experiments: - No gradient accumulation (i.e. gradaccumulationsteps == 1) - shufflebuffersize == 100000 See the OpenVLA GitHub README for instructions on how to run and evaluate this model in the LIBERO simulator.
This model was produced by fine-tuning the OpenVLA 7B model via LoRA (r=32) on the LIBERO-Goal dataset from the LIBERO simulation benchmark. We made a few modifications to the training dataset to improve final performance (see the OpenVLA paper for details). Below are the hyperparameters we used for all LIBERO experiments: - No gradient accumulation (i.e. gradaccumulationsteps == 1) - shufflebuffersize == 100000 See the OpenVLA GitHub README for instructions on how to run and evaluate this model in the LIBERO simulator.
[2026.02.02] Release RynnBrain family weights and inference code. - [2026.02.02] Add cookbooks for cognition, localization, reasoning, and planning. RynnBrain aims to serve as a physics-aware embodied brain: it observes egocentric scenes, grounds language to physical space and time, and supports downstream robotic systems with reliable localization and planning outputs. - Comprehensive egocentric understanding Strong spatial comprehension and egocentric cognition across embodied QA, counting, OCR, and fine-grained video understanding. - Diverse spatiotemporal localization Locates objects, target areas, and predicts trajectories across long episodic context, enabling global spatial…
Model · Image and text to text
swiftvln-satnav-3b-1ep-f32s4-overlap0-pf-h8-pool-s2-noembed
SwiftVLN: training and evaluation code for these checkpoints. - SatNav: satellite-image navigation environments, datasets, and evaluation tools. This checkpoint is designed for SwiftVLN on SatNav, the continuous-state vision-and-language navigation benchmark over satellite imagery. It predicts short navigation-action sequences from an instruction, the current RGB window, and sampled visual memory. - swiftvln-satnav: SwiftVLN trained and evaluated on SatNav - 3b: Qwen2.5-VL 3B backbone - 1ep: trained for one epoch - f32s4: uses a 32-frame window and predicts four actions - overlap0: uses non-overlapping training windows - pf-h8: uses per-frame memory with up to eight history frames…
Also, don't confuse APEX-I-MiniPlus (Standard) with a generic baseline APEX-I-Mini. Traditional APEX-I-Mini drops core experts aggressively to 2-bit IQ2S and leaves output.weight at 3-bit Q3KM, which creates a noticeable perplexity hit on complex reasoning tasks. Standard MiniPlus avoids that degradation floor while keeping boundary layers in linear Q3K for single-cycle vectorized AVX2 CPU dequantization (hitting 23 to 26+ tok/s on DDR4 laptops), while protecting output in Q6K and routers in F32. To put the numbers in perspective: this cuts nearly 2 GB off a flat 3-bit quant (approx. 15.6 GB), and weighs only about approx. 1 GB more than a generic APEX-I-Mini (approx. 12.5 GB). For that…
Expert-paged build of Vontra/Qwen3.8-Flash-Next-MLX-4bit. The weights that are read a fraction at a time live in their own containers, so a machine loads what it needs rather than all Total 105.46 GiB. Of that, 103.94 GiB is the source build, whose bytes moved into containers rather than being copied, and 1.52 GiB is the draft head, which no published build of this model carries. Where the weights fit they are filled from experts.bin and the model runs the stock path at stock speed; where they do not, they stream from disk. Reading the machine decides that, not a flag. To override that: GBXPAGING=off holds the experts resident, GBXPLE=off holds the n-gram table resident. Checked at build…
This repository contains the nvfp4full weight profile of Qwen3.8-27B in the native NInfer.ninfer artifact format, with the z-lab DFlash2 speculative-decoding drafter module added in the upstream W8G32/BF16 format. It is the qwen3.8-27b / nvfp4full identity (same as with the registered DFlash2 module appended — the base tensors are byte-for-byte identical to cometkim v1, and the MTP module is retained (unused and validate-only under --spec dflash2). The base tensors are copied from cometkim v1 byte-for-byte (no re-encoding); only the 66 DFlash2 module objects are produced by the graft tool from the z-lab checkpoint. Verified: source identity qwen3.8-27b/nvfp4full, 1,259 source objects, no…
baidu/Unlimited-OCR as one repository max serve opens directly on Apple Silicon: baidu's weights unchanged (model.safetensors, byte-identical to upstream at revision 07dea832e22aefee32ad281d4b80551282e1c168, sha256 2bc48a7a110061ea58fff65d3169367eebe3aee371ca6968dc2219c1b2855fc6), the tokenizer files as published, and the checkpoint's config.json with two keys removed (automap, modeltype) so MAX can load it without trustremotecode. The runtime is a MAX custom architecture with two Mojo custom ops, served from https://github.com/kthr/unlimited-ocr-max as an OpenAI-compatible endpoint on the Metal GPU or the CPU. The repository also carries model-int8.safetensors, this port's weight-only int8…
(He slides three glasses across the bar—wine for Shakespeare, water for Data, and a mysterious blue liquid for Spock) This model is a merge of: - nightmedia/Qwen3.8-27B-Brainwaves - migtissera/Synthia-4-27B Brainwaves nightmedia/Qwen3.8-27B-Brainwaves migtissera/Synthia-4-27B Late-Night Architectural Takeaways The Quantization Shield (mxfp8 Master Pass): Hitting 0.735 ARC-C on the 8-bit layout proves that your Cold-Fusion flagship anchors and Migel Tissera's Synthia agent paths reached absolute geometric equilibrium. Instead of losing performance to the 0.596 Heretic collapse, the curved hypersphere calculation completely shielded the model's core intelligence. The Perplexity Sweet Spot…
A 27B coding model that runs on one 24 GB NVIDIA card at the full 262,144-token context, with multi-token-prediction speculation and vision still on. It is Swift-Qwen3.8-27B (Qwen3.8-27B post-trained by UkisAI for shorter reasoning), requantized with a coding-calibrated importance matrix. Nothing was trained here: this is a quantization and a deployment recipe, measured end to end. IQ4XS/Potato-CODER-24GB-IQ4XS.gguf (the model, MTP head embedded), mmproj-Potato-CODER-24GB-Q80.gguf (vision projector, Q80 of UkisAI's F16 projector; the vision tower is unchanged from Qwen3.8-27B), LICENSE (Swift Open License), LICENSE-APACHE-2.0 (base model), NOTICE, SHA256SUMS. Use our llama.cpp branch…
This model is a fine-tuned version of Qwen/Qwen3.5-4B-Base on the Omni-Edu-70K dataset. The following hyperparameters were used during training: - learningrate: 5e-06 - trainbatchsize: 1 - evalbatchsize: 8 - distributedtype: multi-GPU - numdevices: 8 - gradientaccumulationsteps: 8 - totaltrainbatchsize: 64 - totalevalbatchsize: 64 - lrschedulertype: cosine - lrschedulerwarmupsteps: 0.1 - numepochs: 3.0 - Transformers 5.2.0 - Pytorch 2.10.0 - Datasets 4.0.0 - Tokenizers 0.22.2
This model is a fine-tuned version of Qwen/Qwen3.5-9B-Base on the Omni-Edu-70K dataset. The following hyperparameters were used during training: - learningrate: 5e-06 - trainbatchsize: 1 - evalbatchsize: 8 - distributedtype: multi-GPU - numdevices: 8 - gradientaccumulationsteps: 8 - totaltrainbatchsize: 64 - totalevalbatchsize: 64 - lrschedulertype: cosine - lrschedulerwarmupsteps: 0.1 - numepochs: 3.0 - Transformers 5.2.0 - Pytorch 2.10.0 - Datasets 4.0.0 - Tokenizers 0.22.2
This is a merge of pre-trained language models created using mergekit. This model was merged using the aura merge method. Aura is an experimental method with a live heatmap visualizer. This model took 10 hours to merge using graphv18.py The following models were included in the merge: - TheDrummer/Orion-26B-A4B-v1.1 - Gryphe/Pantheon-Reasoning-26B-A4B-1.1-V2 - electroglyph/gemma4-26b-fiction-bf16 The following YAML configuration was used to produce this model
licensename: swift-open-license-1.0 licenselink: https://huggingface.co/ukisai/Swift-Qwen3.8-27b/blob/main/LICENSE libraryname: transformers pipelinetag: image-text-to-text - qwen38 - efficient-thinking - reasoning - token-efficient - genesis - lora basemodel: ukisai/Swift-Qwen3.8-27b basemodelrelation: finetune
Questions
Which Image and text to text models are most downloaded?
By monthly downloads reported by the Hugging Face Hub: Gemma-4-E4B-Uncensored-HauhauCS-Aggressive (2M); Qwen2.5-VL-7B-Instruct-AWQ (2M); gemma-4-26B-A4B-it-AWQ-4bit (1.8M).