Leaderboard / Google
Gemma 3 270M
Google · Gemma 3 · released 2025-08-14 · data checked 2026-11-01
Gemma 3 270M is a 270M-parameter dense model from Google (United States), released 2025-08-14 under the Gemma license. It doesn't have enough reported benchmarks for an index score yet. It needs about 0.7 GB of memory at 4-bit with an 8K-token context.
Meant to be fine-tuned for narrow tasks, not used as a general assistant.
Benchmarks
Each bar shows this model's score. The tick marks the best tiny model on that benchmark.
No benchmark results for this release yet. Know of one? Send it.
Specifications
- Lab
- Google (US · United States)
- Architecture
- Dense transformer
- Total parameters
- 270M
- Active parameters
- All (dense)
- Size class
- Tiny
- Context window
- 32K tokens
- Input modalities
- text
- Reasoning
- Instruct (no thinking)
- License
- Gemma
- Openness
- Open weights
- Memory to run
- ~0.7 GB at 4-bit, 8K context (more sizes)
- Hugging Face
- 68.3K downloads (30 days) · 703 likes
Compare with
- Gemma 3 270M vs LFM2.5 350M (350M, index 12.6)
- Gemma 3 270M vs Qwen3.5 0.8B (900M, index 10.7)
- Gemma 3 270M vs Gemma 3 1B (1B, index 6.1)
Sources
Something wrong here? Report it.
Run it locally
Uses unsloth/gemma-3-270m-it-GGUF (Q4_K_M). Swap the tag for another size: F16, IQ4_NL, IQ4_XS, Q2_K, Q2_K_L, Q3_K_M, Q3_K_S, Q4_0….
ollama run hf.co/unsloth/gemma-3-270m-it-GGUF:Q4_K_Mllama-server -hf unsloth/gemma-3-270m-it-GGUF:Q4_K_Mpip install -U mlx-lm
mlx_lm.chat --model mlx-community/gemma-3-270m-it-4bitvllm serve google/gemma-3-270m-itMemory needed: weights, context cache and runtime buffers.
| Quantization | 4K context | 32K context |
|---|---|---|
| 16-bit F16 | 1.1 GB | 1.2 GB |
| 8-bit Q8_0 | 0.8 GB | 0.9 GB |
| 4-bit Q4_K_M | 0.7 GB | 0.8 GB |