Chat
Gemma 3 27B
Practical Gemma-family model still worth testing locally, now positioned as a useful baseline behind Gemma 4.
Google · Gemma
Editorial review
Model checkpoints, context windows, provider support, local runtime compatibility, and license terms can change quickly. Verify the exact model card before production or commercial use.
Best for
Developers testing capable medium-sized chat models with broad tooling support.
Who should use it
- Developers testing capable medium-sized chat models with broad tooling support.
- Builders who want local or self-hosted testing options.
Common workflows
- General chat, multilingual, local and hosted apps
- chat workflows
- local workflows
- legacy workflows
Deployment and hardware notes
Quantized builds can fit higher-end consumer GPUs or unified-memory Macs.
License and usage notes
Gemma Terms of Use. Open weights. Verify the exact model card and license terms for the checkpoint or hosted provider you use.
Strengths
- Open weights model option for Gemma workflows.
- Developers testing capable medium-sized chat models with broad tooling support.
- Can be tested locally with quantized builds on higher-end consumer GPUs or unified-memory systems.
- Tracked as Practical local in the OpenSourcesAI model directory.
Limitations
- License is custom; newer Gemma models may be more relevant for current benchmarking.
- Quantized builds can fit higher-end consumer GPUs or unified-memory Macs.
- Context window and limits: 131,072 tokens.
- Verify the exact model card, provider docs, license, and serving support before production use.
Local workflow notes
Can be tested locally with quantized builds on higher-end consumer GPUs or unified-memory systems.
Local runtimes: Ollama, LM Studio, llama.cpp, Transformers
Platforms: Windows, macOS, Linux
Will Gemma 3 27B run on your machine?
Gemma 3 27B is 27B parameters and needs 18.5 GB of VRAM at Q4_K_M — 17 GB of weights plus 1.5 GB of runtime overhead for the inference server itself.
VRAM by quantization
| Quantization | Weights | Needs (with overhead) | Quality |
|---|---|---|---|
| Q4_K_M | 17 GB | 18.5 GB | good |
| Q8_0 | 29 GB | 30.5 GB | high |
| FP16 | 56 GB | 57.5 GB | reference |
Fit on common hardware at Q4_K_M
| Hardware | Memory the model can use | System RAM | Verdict |
|---|---|---|---|
| CPU Only | None (CPU only) | 16 GB | Too large |
| RTX 4060 Laptop | 8 GB | 16 GB | Too large |
| RTX 3060 (12GB) | 12 GB | 32 GB | CPU offload |
| RTX 4060 Ti (16GB) | 16 GB | 32 GB | CPU offload |
| RTX 3090 | 24 GB | 64 GB | Comfortable |
| Apple Silicon (Unified Memory) 36 GB | 27 GB of 36 GB | 36 GB | Comfortable |
| RTX 5090 | 32 GB | 64 GB | Comfortable |
Comfortable means VRAM clears the requirement by 2 GB or more. Tight means it covers the requirement with no margin. CPU offload means the model does not fit in VRAM but system RAM is at least 1.6× the weights, so it will run at reduced speed — expect roughly 1–5 tokens per second. Figures are weights plus a fixed runtime overhead and exclude KV-cache growth, which scales with context length.
Apple Silicon shares one pool of memory between the system and the GPU, so a model cannot use all of it. These rows apply the same 75% usable fraction the Compatibility Checker uses, which is why a 36 GB Mac is graded on less than 36 GB.
VRAM fit by quantization level
Enter your GPU VRAM below to see which quantization of Gemma 3 27B fits and get the Ollama run command.
Sources to verify
Related resources
Continue with model source notes, local tools, and implementation guides related to this model.
Model ecosystem connections
Use these next-step links to move from this profile into related tools, comparisons, guides, stacks, and curated shortlists.
Compatible hardware
Recommended runtimes and tools
Setup and deployment
Related model pages
Guides, stacks, and comparisons
Ready to run this model locally?
Find a compatible interface in our Local AI Tools directory →