Gemma 4 E2B
- In Input modalities: TextImageAudio Out Output modalities: Text
- Reasoning
- Reasoning can be off
- Tool use
- Open weights
Prices
0 offerings · prices as listedNo priced offering is listed for this model yet.
Details
- Licence
- apache-2.0
- Family
- gemma-e2b
Not stated on Google’s pages: max input, reasoning efforts, max thinking budget, structured output, built-in tools, knowledge cutoff, aliases.
Run locally
Weights: google/gemma-4-E2B- Parameters 5.1B
- Dense
- Native precision BF16
- Layers 35
- KV cache per token 35 KB
- Config positions 128K (config.json max_position_embeddings)
- API context 125K (maker’s API)
- Needs
- ≈ 10.1 GB (weights 9.5 (maker’s file) + KV cache 0.1 + overhead 0.5)
- Usable
- 21.6 GB (90% of its memory)
- Generation speed
- 49–73 tok/s
- Longest context
- 125K (the model’s limit)
- First token
- about 0.17–0.25 s for a 2K prompt
| Precision | Weights | Needs at 8K | On your hardware | Longest context |
|---|---|---|---|---|
| BF16 (native)maker’s file | 9.5 GB | ≈ 10.1 | ●Runs well 49–73 tok/s | 125K |
| Q8_0 | 5.1 GB | ≈ 5.6 | ●Runs well 91–137 tok/s | 125K |
| FP8 | 4.8 GB | ≈ 5.3 | ●Runs well 97–145 tok/s | 125K |
| Q6_K | 3.9 GB | ≈ 4.5 | ●Runs well 118–177 tok/s | 125K |
| Q5_K_M | 3.4 GB | ≈ 4.0 | ●Runs well 136–203 tok/s | 125K |
| Q5_0 | 3.3 GB | ≈ 3.8 | ●Runs well 140–210 tok/s | 125K |
| Q4_K_M | 2.9 GB | ≈ 3.5 | ●Runs well 158–237 tok/s | 125K |
| Q4_0 | 2.7 GB | ≈ 3.3 | ●Runs well 171–256 tok/s | 125K |
| IQ4_XS | 2.5 GB | ≈ 3.1 | ●Runs well 180–270 tok/s | 125K |
| Q3_K_Mquality drops | 2.3 GB | ≈ 2.9 | ●Runs well 196–293 tok/s | 125K |
| IQ3_XXSquality drops | 1.8 GB | ≈ 2.4 | ●Runs well 248–372 tok/s | 125K |
| Q2_Kquality drops | 1.8 GB | ≈ 2.4 | ●Runs well 253–379 tok/s | 125K |
| IQ2_XXSquality drops | 1.2 GB | ≈ 1.8 | ●Runs well 362–543 tok/s | 125K |
Click a row to choose a precision. “Maker’s file” sizes are the maker’s own files; others are parameters × bits per weight.
Compared with other hardware
Each at the highest precision it runs. Click one for details.
Smallest that runs it (Q4_K_M, 8K): GPU GeForce RTX 3080; unified memory Apple M5 · 16 GB.
Run commands
vllm serve google/gemma-4-E2B --max-model-len 8192
Memory to fine-tune
Batch 1, 2,048-token sequences, gradient checkpointing; LoRA and QLoRA train rank-16 adapters on the attention projections.
Estimates; facts from the maker’s Hugging Face repositories, as of 2026-11-06. How it’s estimated
Cited scores
From third-party leaderboards, not measured by usThis model isn’t on any leaderboard we cite yet.
Changes
All changes →No changes recorded yet.
Sources
- Gemma 4 model card (opens in a new tab) ai.google.dev · checked Nov 1, 2026
- google/gemma-4-E2B model card (opens in a new tab) huggingface.co · checked Oct 30, 2026
- google/gemma-4-E2B config.json (opens in a new tab) huggingface.co · checked Nov 4, 2026
- google/gemma-4-E2B files (opens in a new tab) huggingface.co · checked Nov 1, 2026
- google/gemma-4-E2B model card (opens in a new tab) huggingface.co · checked Nov 1, 2026
Google’s official policies: data use, privacy, terms →
Some values on this page are still waiting to be checked against official pages: a filled dot is verified, a hollow one is not yet (with or without a source on file). How we verify data
Google · United States · ai.google.dev
Related models
Snapshots, the same family and other models from this provider