{"slug": "google-deepmind-releases-gemma-4-12b-an-encoder-free-multimodal-model-with-audio", "title": "Google DeepMind Releases Gemma 4 12B: An Encoder-Free Multimodal Model with Native audio that runs on a 16 GB laptop", "summary": "Google DeepMind released Gemma 4 12B, an encoder-free multimodal model that processes text, images, audio, and video natively without separate vision or audio encoders. The 12-billion-parameter model runs agentic workflows on consumer laptops with 16 GB of RAM and is available under the Apache 2.0 license. The design eliminates latency from traditional encoders, allowing the LLM backbone to process inputs immediately while enabling unified fine-tuning across all modalities.", "body_md": "Google DeepMind just released [Gemma 4 12B](https://huggingface.co/collections/google/gemma-4), a dense multimodal model that strips out traditional encoders entirely. Vision and audio flow straight into the LLM backbone. The result is a model that runs agentic workflows on a consumer laptop with 16 GB of RAM. It ships under the Apache 2.0 license.\n\n**Model Overview & Access**\n\nGemma 4 12B is a 12-billion-parameter decoder-only transformer. It handles text, images, audio, and video natively. There are no separate vision or audio encoders. The decoder uses the same structure as the Gemma 4 31B Dense model. It bridges the gap between the edge-friendly E4B and the larger 26B Mixture of Experts variant.\n\n**Architecture:** Unified, encoder-free decoder-only transformer.**Modalities:** Text, image, video, and native audio input — the first mid-sized Gemma with audio.**Hardware requirement:** 16 GB VRAM or unified memory. Runs on consumer GPU laptops and Apple Silicon Macs.**License:** Apache 2.0. Weights are open and publicly downloadable.**Inference stack:** Compatible with llama.cpp, MLX, vLLM, Ollama, SGLang, Unsloth, and LM Studio.**Download:**[Hugging Face](https://huggingface.co/collections/google/gemma-4)and[Kaggle](https://www.kaggle.com/models/google/gemma-4). The instruct variant is`google/gemma-4-12B-it`\n\n.**Integration:** Hugging Face Transformers, LiteRT-LM CLI, and an OpenAI-compatible local API server via`litert-lm serve`\n\n.\n\nA dedicated Multi-Token Prediction (MTP) drafter model is also released. It reduces inference latency on local hardware.\n\n**Architecture: The Encoder-Free Design**\n\nEvery prior mid-sized Gemma model used separate Transformer encoders for vision and audio. Those encoders added latency and parameter overhead. The medium-sized Gemma 4 models carry a 550M-parameter vision encoder. The E2B and E4B models include a 300M-parameter audio encoder. All of that is gone in the 12B.\n\n**Vision embedder (35M parameters**): Raw images are split into 48×48 pixel patches. Each patch is projected to the LLM’s hidden dimension with a single matrix multiplication. There is no attention layer; each patch is processed independently. Spatial position is injected using a factorized coordinate lookup: a learned X matrix and a learned Y matrix. For a patch at (x, y), the model looks up two learned embeddings and adds them to form a position vector. This is added to the patch embedding, followed by normalization. That is the entire vision pipeline.\n\n**Audio wave projection**: Raw 16 kHz audio is sliced into 40 ms frames. Each frame contains 640 values. Those values are linearly projected into the same embedding space as text tokens. There is no feature extraction and no conformer layers. The LLM’s existing Rotary Position Embedding (RoPE) handles the 1-D temporal sequence. The audio encoder in the E2B and E4B used 12 conformer layers. All of that is removed.\n\n**Importance:** The unified weight space means you no longer co-tune separate frozen encoders. Downstream fine-tuning with LoRA or full tuning updates vision, audio, and text processing in a single pass. Hugging Face Transformers and Unsloth already support this.\n\nThe encoder-free design reduces multimodal latency. The LLM backbone starts processing immediately. No encoder must finish first.\n\n**Capabilities & Performance**\n\nGoogle DeepMind team has not published full benchmark results in the initial launch materials. The official release notes state the 12B model performs nearing the 26B MoE model on standard benchmarks, at less than half the total memory footprint.\n\nThe model’s demonstrated capabilities include:\n\n**Automatic speech recognition.** Transcribes audio natively without an external ASR pipeline.**Agentic reasoning.** Runs multi-step workflows locally, with performance approaching the 26B MoE model.**Diarization.** Distinguishes speakers in audio input.**Video understanding.** Processes video frames alongside audio. A demo analyzed a 5-minute Google I/O keynote segment using 313 frames at 1 FPS with a visual token budget of 70 per frame.**Coding.** Built a Gradio image-processing app using its own code generation, served locally with llama.cpp.**Multimodal agentic workflows.** The official Gemma Skills repository at[github.com/google-gemma/gemma-skills](https://github.com/google-gemma/gemma-skills)provides pre-built agent capabilities.\n\nIn Google’s own Google AI Edge Eloquent app, the switch to Gemma 4 12B produced what Google reports as a 60%+ jump in overall quality, with improved instruction following and scope adherence.\n\n**Marktechpost’s Visual Explainer**\n\n[marktechpost.com](https://www.marktechpost.com)\n\n**Key Takeaways**\n\n- Google DeepMind released Gemma 4 12B, a dense encoder-free multimodal model under the Apache 2.0 license.\n- Vision and audio feed straight into the LLM backbone — no separate vision (550M) or audio (300M) encoders.\n- A 35M vision embedder uses a single matmul plus factorized X/Y position lookup; audio projects raw 16 kHz frames directly.\n- It is the first mid-sized Gemma with native audio, and adds video, running on a 16 GB laptop.\n- Benchmark performance nears the 26B MoE model at less than half the memory footprint.\n\nCheck out the ** Model Weights** and\n\n**Also, feel free to follow us on**\n\n[Technical details](https://blog.google/innovation-and-ai/technology/developers-tools/introducing-gemma-4-12b/).**and don’t forget to join our**[Twitter](https://x.com/intent/follow?screen_name=marktechpost)\n\n**and Subscribe to**\n\n[150k+ ML SubReddit](https://www.reddit.com/r/machinelearningnews/)**. Wait! are you on telegram?**\n\n[our Newsletter](https://www.aidevsignals.com/)\n\n[now you can join us on telegram as well.](https://t.me/machinelearningresearchnews)Need to partner with us for promoting your GitHub Repo OR Hugging Face Page OR Product Release OR Webinar etc.? [Connect with us](https://forms.gle/wbash1wF6efRj8G58)", "url": "https://wpnews.pro/news/google-deepmind-releases-gemma-4-12b-an-encoder-free-multimodal-model-with-audio", "canonical_source": "https://www.marktechpost.com/2026/06/03/google-deepmind-releases-gemma-4-12b-an-encoder-free-multimodal-model-with-native-audio-that-runs-on-a-16-gb-laptop/", "published_at": "2026-06-03 18:46:41+00:00", "updated_at": "2026-06-03 19:07:07.564939+00:00", "lang": "en", "topics": ["artificial-intelligence", "large-language-models", "generative-ai", "ai-products", "ai-tools"], "entities": ["Google DeepMind", "Gemma 4 12B", "Apache 2.0", "Hugging Face", "Kaggle", "llama.cpp", "MLX", "vLLM"], "alternates": {"html": "https://wpnews.pro/news/google-deepmind-releases-gemma-4-12b-an-encoder-free-multimodal-model-with-audio", "markdown": "https://wpnews.pro/news/google-deepmind-releases-gemma-4-12b-an-encoder-free-multimodal-model-with-audio.md", "text": "https://wpnews.pro/news/google-deepmind-releases-gemma-4-12b-an-encoder-free-multimodal-model-with-audio.txt", "jsonld": "https://wpnews.pro/news/google-deepmind-releases-gemma-4-12b-an-encoder-free-multimodal-model-with-audio.jsonld"}}