# Gemma 4: Google DeepMind bets on open models for the edge Three variants tuned for compute and memory efficiency, reasoning, and agentic flows. Open and enterprise-ready stopped being opposites. Google DeepMind published the **Gemma 4** family in three variants: **2B, 12B, and 27B parameters**. This is not another incremental release. The benchmarks show the 27B model beating Llama 3 70B on reasoning tasks (MMLU, HumanEval) with less than half the parameters. The 12B holds its own against Mistral 7B×8 (Mixtral) on code generation while using a fraction of the memory. The message is clear: intelligence per parameter is now a first-class metric. The 2B model is the one to watch if you work at the edge. It runs on smartphones with 4 GB of RAM, on a Raspberry Pi 5, and on industrial gateways. Until yesterday, the only option in those scenarios was to send everything to a cloud API and hope latency did not kill the experience. Now you can run text classification, entity extraction, and even simple conversational flows on the device itself. ## Open vs. closed: the real trade-off Open models are not charity. Google knows that the more people train and deploy Gemma, the more its architecture becomes the de facto standard — and the more use cases there are to feed the next versions. That does not invalidate the offer. It means you need to understand the game before you join it. The Gemma license allows commercial use, fine-tuning, and redistribution with attribution. Compared to Llama 3, which restricts applications with more than 700 million monthly users, in practice it is more permissive for most companies. Compared to Mistral, which moved to more restrictive licenses on its larger models, Gemma stays consistent across every variant. The part nobody talks about: an open model is not an audited model. You get the weights, the training documentation, and a model card. You do not get a guarantee that the dataset holds no European personal data, that bias in Portuguese was tested rigorously, or that the next version will not break your fine-tuning pipeline. That responsibility is yours. ## What it means for enterprise **Fine-tuning and cost.** Gemma 4 uses an attention architecture that makes LoRA efficient on commodity hardware. A supervised fine-tune of the 12B model runs on a single A100 80 GB in under 4 hours for datasets of 50,000 examples. If you are used to reserving entire clusters just to train adapters for larger models, that changes the cost equation. **Data sovereignty.** With the 2B or the 12B running on-prem, sensitive data never leaves the perimeter. In healthcare, finance, and government, that is a prerequisite, not a nice-to-have. Gemma 4 makes it viable to run a language model inside your own datacenter without large-scale GPU infrastructure. **Model supply chain.** Every model that enters your stack is a dependency. It needs versioning, vulnerability scanning on the inference container, drift monitoring, and a rollback plan. Companies that treat a model as a disposable artifact will find out at the worst possible moment that they have no reproducibility. **Agents and orchestration.** Gemma 4 27B was designed with native support for tool-calling and multi-step flows. On agentic benchmarks (GAIA, ToolBench) it beats earlier open models by a significant margin. That makes it practical to build architectures where the local model decides which tools to call, executes, evaluates the result, and iterates, all without a round trip to an external API. ## Where Uranus sees value We work with clients who need AI running in production, not in demos. What changes with Gemma 4: - **Hybrid architecture with a purpose.** Sensitive workloads run the 12B at the perimeter. Tasks that need maximum capability go up to the 27B in a private cloud, with unified orchestration and observability. This is not multi-model for the sake of it. Different workloads have different latency, cost, and compliance requirements. - **Business agents that work.** We combine RAG, tool-calling, and explicit security policies. The model is one layer. The value is in engineering the whole system: how the agent decides, how it fails, how it scales, and how it is monitored. - **Governance as engineering.** Licensing, attribution, a model version registry, reproducible inference, and output auditing. Calling that bureaucracy is what keeps a project stuck in the prototype phase, especially in a regulated sector. The open model market is maturing fast, and the competition between Gemma, Llama, and Mistral benefits everyone building on those foundations. But models are a commodity. The winners are the ones who integrate model, data, and processes with the same rigor as any critical system. That is the kind of work we do in [AI agents](/capabilities#agents) and [software engineering](/capabilities#engineering). --- **Read also:** [Claude Mythos Preview and AI-assisted cybersecurity](/blog/claude-mythos-preview-ai-assisted-cybersecurity) · [Meta introduces Muse Spark](/blog/meta-muse-spark-superintelligence-team) More detail on the [official Gemma page at Google DeepMind](https://deepmind.google/models/gemma/).