# Qwen2.5-Omni-3B by Qwen: Open-Weight Model Source: https://savrn.com/models/qwen2-5-omni-3b Markdown alternate of the page above; the site index is https://savrn.com/llms.txt --- ## Runs On What it takes to serve Qwen2.5-Omni-3B (5.5B parameters): the memory its weights need at each precision, and the cheapest way to rent enough data-center GPUs to hold them. | Precision | Weights | Memory needed | Cheapest setup | Per hour | Also fits | | --- | --- | --- | --- | --- | --- | | 16-bit | 11.1 GB | 13.3 GB | 1x [MI300X](https://savrn.com/ai-index/pricing/gpus/mi300x) (192 GB) Vultr | $1.85 | [1x H100](https://savrn.com/ai-index/pricing/gpus/h100) $1.99 · [1x MI325X](https://savrn.com/ai-index/pricing/gpus/mi325x) $2.00 | | 8-bit | 5.5 GB | 6.6 GB | 1x [MI300X](https://savrn.com/ai-index/pricing/gpus/mi300x) (192 GB) Vultr | $1.85 | [1x H100](https://savrn.com/ai-index/pricing/gpus/h100) $1.99 · [1x MI325X](https://savrn.com/ai-index/pricing/gpus/mi325x) $2.00 | | 4-bit | 2.8 GB | 3.3 GB | 1x [MI300X](https://savrn.com/ai-index/pricing/gpus/mi300x) (192 GB) Vultr | $1.85 | [1x H100](https://savrn.com/ai-index/pricing/gpus/h100) $1.99 · [1x MI325X](https://savrn.com/ai-index/pricing/gpus/mi325x) $2.00 | Memory is the weights at that precision plus 20% for the runtime and a short context; a long context needs more. Prices are the lowest on-demand hourly rates in the [SAVRN Index](https://savrn.com/ai-index/pricing/gpus), read Oct 24, 2026. [Qwen2.5-Omni-3B on every accelerator the SAVRN Index prices, at every precision](https://savrn.com/models/qwen2-5-omni-3b/gpus) ## SAVRN's Notes on Qwen2.5-Omni-3B We file this one under whole voice assistant in a single checkpoint. Qwen2.5-Omni-3B takes text, images, audio and video and answers in text and streamed speech. The 3B in the name undersells it: the file holds 5.5 billion parameters in bfloat16, 12 GB across 18 files. Memory in use is 13.3 GB at 16-bit, 6.6 GB at 8-bit and 3.3 GB at 4-bit, so even full precision fits on nearly any accelerator you already own. The cheapest rental we track is one MI300X at $1.85 an hour, which is paying for 192 GB to use 13. Two things to settle first. The license is recorded as other with no summary, so put the publisher's terms in front of counsel before a commercial speech product. And no context length is on record, so measure how much audio or video one request holds before promising a session length. The weights date from April 30, 2025 with no update since; the paper is arXiv:2503.20215. SAVRN Research, 2026-09-18 ## Model Card Qwen2.5-Omni is an end-to-end multimodal model designed to perceive diverse modalities, including text, images, audio, and video, while simultaneously generating text and natural speech responses in a streaming manner. We conducted a comprehensive evaluation of Qwen2.5-Omni, which demonstrates strong performance across all modalities when compared to similarly sized single-modality models and closed-source models like Qwen2.5-VL-7B, Qwen2-Audio, and Gemini-1.5-pro. In tasks requiring the integration of multiple modalities, such as OmniBench, Qwen2.5-Omni achieves state-of-the-art performance. Furthermore, in single-modality tasks, it excels in areas including speech recognition (Common… Excerpt from the card by Qwen, licensed other. ## Configuration Architecture Qwen2_5OmniModel Stored precision bfloat16 Model type qwen2_5_omni ## Identity and Version Repository Qwen/Qwen2.5-Omni-3B Publisher Qwen Task Any to any Modality Multimodal Library transformers Parameters 5.5B parameters Languages en Revision f75b40e3da2003cdd6e1829b1f420ca70797c34e First published 2025-04-30 Last updated 2025-04-30 ## Files and Weights 18 files, 12.0 GB in total. The weights are 4 files totalling 12.0 GB in pt, safetensors. Weights4 files · 12.0 GB Configuration7 files · 258.5 KB Tokenizer4 files · 15.9 MB Documentation2 files · 44.2 KB Repository1 file · 1.5 KB Every file | File | Type | Size | SHA-256 | | --- | --- | --- | --- | | model-00001-of-00003.safetensors | Weights | 5.0 GB | 349972cebff4 | | model-00002-of-00003.safetensors | Weights | 5.0 GB | b29a76dfefb3 | | model-00003-of-00003.safetensors | Weights | 2.0 GB | 8f2138170a87 | | spk_dict.pt | Weights | 259.5 KB | 6a05609b28f5 | | added_tokens.json | Configuration | 579 B | — | | chat_template.json | Configuration | 1.3 KB | — | | config.json | Configuration | 13.2 KB | — | | generation_config.json | Configuration | 74 B | — | | model.safetensors.index.json | Configuration | 241.8 KB | — | | preprocessor_config.json | Configuration | 667 B | — | | special_tokens_map.json | Configuration | 832 B | — | | LICENSE | Documentation | 7.4 KB | — | | README.md | Documentation | 36.8 KB | — | | .gitattributes | Repository | 1.5 KB | — | | merges.txt | Tokenizer | 1.7 MB | — | | tokenizer.json | Tokenizer | 11.4 MB | 8441917e39ae | | tokenizer_config.json | Tokenizer | 6.5 KB | — | | vocab.json | Tokenizer | 2.8 MB | — | ## License and Download License other Access Open weights, no gate Download size 12.0 GB [Download from Qwen](https://modelscope.cn/models/Qwen/Qwen2.5-Omni-3B) Released by Qwen through ModelScope. ## Built From - Described by [arXiv:2503.20215](https://savrn.com/papers/qwen2-5-omni-technical-report) ## Memory Requirements | Precision | Weights in memory | | --- | --- | | As published | 12.0 GB | | 16-bit | 11.1 GB | | 8-bit | 5.5 GB | | 4-bit | 2.8 GB | Weights only, from the published parameter count; the key-value cache and runtime add to this. ## Built on This Model - Derived from[jevtrainer-omni3b-2026-10-04](https://savrn.com/models/jevtrainer-omni3b-2026-10-04) - Quantized from[RK182X-MLLM-Qwen2.5-Omni-3B](https://savrn.com/models/rk182x-mllm-qwen2-5-omni-3b) - Derived from[RK182X-MLLM-Qwen2.5-Omni-3B](https://savrn.com/models/rk182x-mllm-qwen2-5-omni-3b) ## Compare Qwen2.5-Omni-3B - [gemma-4-E2B-it vs Qwen2.5-Omni-3B](https://savrn.com/models/compare/gemma-4-e2b-it-vs-qwen2-5-omni-3b) ## Questions About Qwen2.5-Omni-3B ### How much GPU memory does Qwen2.5-Omni-3B need? About 13.3 GB at 16-bit and 3.3 GB at 4-bit: the weights (5.5B parameters) plus a working margin. A long context needs more. ### What is the cheapest GPU to run Qwen2.5-Omni-3B on? At 16-bit, 1x MI300X from $1.85 an hour; at 4-bit, 1x MI300X from $1.85 an hour, at the lowest on-demand prices the SAVRN Index lists. ### What license is Qwen2.5-Omni-3B released under? other, as its publisher declares it. Read the license text before commercial use. ## Similar Models Model · Any to any ### [gemma-4-E2B-it-qat-w4a16-ct](https://savrn.com/models/gemma-4-e2b-it-qat-w4a16-ct) [Google](https://savrn.com/model-publishers/google) 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… Open weights apache-2.0 5.6B parameters 131,072 tokens transformers [View model](https://savrn.com/models/gemma-4-e2b-it-qat-w4a16-ct) Model · Any to any ### [gemma-4-E2B-it-AWQ-INT4](https://savrn.com/models/gemma-4-e2b-it-awq-int4) [Cyankiwi](https://savrn.com/model-publishers/cyankiwi) 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… Open weights apache-2.0 5.2B parameters 131,072 tokens transformers [View model](https://savrn.com/models/gemma-4-e2b-it-awq-int4) Model · Any to any ### [gemma-4-E2B-it](https://savrn.com/models/gemma-4-e2b-it) [Google](https://savrn.com/model-publishers/google) 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… Open weights apache-2.0 5.1B parameters 131,072 tokens transformers [View model](https://savrn.com/models/gemma-4-e2b-it) Model · Any to any ### [gemma-4-E2B-it-MLX-4bit](https://savrn.com/models/gemma-4-e2b-it-mlx-4bit) [LM Studio Community](https://savrn.com/model-publishers/lmstudio-community) LM Studio Community models highlights program. Highlighting new & noteworthy models by the community. Join the conversation on Discord. 4-bit quantized version of gemma-4-E2B-it using MLX, optimized for Apple Silicon. Special thanks to the Apple Machine Learning Research team for creating MLX. LM Studio is not the creator, originator, or owner of any Model featured in the Community Model Program. Each Community Model is created and provided by third parties. LM Studio does not endorse, support, represent or guarantee the completeness, truthfulness, accuracy, or reliability of any Community Model. You understand that Community Models can produce content that might be offensive, harmful… Open weights apache-2.0 5.1B parameters 131,072 tokens transformers [View model](https://savrn.com/models/gemma-4-e2b-it-mlx-4bit) Model · Any to any ### [gemma-4-E2B-it-MLX-8bit](https://savrn.com/models/gemma-4-e2b-it-mlx-8bit) [LM Studio Community](https://savrn.com/model-publishers/lmstudio-community) LM Studio Community models highlights program. Highlighting new & noteworthy models by the community. Join the conversation on Discord. 8-bit quantized version of gemma-4-E2B-it using MLX, optimized for Apple Silicon. Special thanks to the Apple Machine Learning Research team for creating MLX. LM Studio is not the creator, originator, or owner of any Model featured in the Community Model Program. Each Community Model is created and provided by third parties. LM Studio does not endorse, support, represent or guarantee the completeness, truthfulness, accuracy, or reliability of any Community Model. You understand that Community Models can produce content that might be offensive, harmful… Open weights apache-2.0 5.1B parameters 131,072 tokens transformers [View model](https://savrn.com/models/gemma-4-e2b-it-mlx-8bit) Model · Any to any ### [gemma-4-E2B-it-MLX-6bit](https://savrn.com/models/gemma-4-e2b-it-mlx-6bit) [LM Studio Community](https://savrn.com/model-publishers/lmstudio-community) LM Studio Community models highlights program. Highlighting new & noteworthy models by the community. Join the conversation on Discord. 6-bit quantized version of gemma-4-E2B-it using MLX, optimized for Apple Silicon. Special thanks to the Apple Machine Learning Research team for creating MLX. LM Studio is not the creator, originator, or owner of any Model featured in the Community Model Program. Each Community Model is created and provided by third parties. LM Studio does not endorse, support, represent or guarantee the completeness, truthfulness, accuracy, or reliability of any Community Model. You understand that Community Models can produce content that might be offensive, harmful… Open weights apache-2.0 5.1B parameters 131,072 tokens transformers [View model](https://savrn.com/models/gemma-4-e2b-it-mlx-6bit) ## Qwen [All models and datasets](https://savrn.com/model-publishers/qwen) ## Versions - [f75b40e3da20](https://savrn.com/models/qwen2-5-omni-3b/versions/f75b40e3da20) · current 2026-09-18 ## Explore More - [All any to any models](https://savrn.com/models/tasks/any-to-any) - [Model comparisons](https://savrn.com/models/comparisons) - [The model directory](https://savrn.com/models) - [Open model prices by host](https://savrn.com/ai-index/pricing/open-models) ## Source - Repository metadata, read 2026-10-20. - [Qwen website](https://qwen.ai/) - [Hugging Face record](https://huggingface.co/Qwen/Qwen2.5-Omni-3B) - [How the hub is built](https://savrn.com/model-hub/methodology)