gemma-4-E4B-it

reasoning
multimodal

Gemma‑4‑E4B‑it is the second largest dense model in the new family of open LLMs from Google, offering solid performance with extremely economical resource consumption. The model uses the innovative Per‑Layer Embeddings (PLE) technique, which fundamentally changes the approach to building small language models. In standard transformers, each token receives a single embedding vector that passes through all network layers. PLE works differently: each of the 42 decoder layers gets its own small embedding per token. These embeddings are stored in large tables (the total model size reaches 8 billion parameters), but during inference only the effective part — 4.5 billion — is active.

This architecture allows the E4B model to achieve performance comparable to models two to three times larger. According to community feedback, E4B confidently surpasses Gemma‑3 27B in several tasks, even though its effective size is 12 times smaller. The model supports a context window of 128 thousand tokens and uses hybrid attention with a sliding window of 512 tokens. A key difference between E4B and the larger 31B model is built‑in audio support (an encoder of ~300M parameters), which makes the model universal — it can simultaneously process text, images, and sound.

Developers position E4B as a model for complex local tasks. It is ideally suited for use on high‑performance laptops, powerful mobile devices, and embedded systems. Thanks to the Apache 2.0 licence, the model can be freely fine‑tuned and integrated into commercial products that operate under tight memory constraints.

For the developers’ usage recommendations for the model, please refer to this link: https://ai.google.dev/gemma/docs/core/model_card_4?hl=en


Announce Date: 02.03.2026
Parameters: 8B
Context: 132K
Layers: 42, using full attention: 7
Attention Type: Sliding Window Attention
Developer: Google DeepMind
Transformers Version: 5.5.0.dev0
License: Apache 2.0

Public endpoint

Use our pre-built public endpoints for free to test inference and explore gemma-4-E4B-it capabilities. You can obtain an API access token on the token management page after registration and verification.
Model Name Context Type GPU Status Link
There are no public endpoints for this model yet.

Private server

Rent your own physically dedicated instance with hourly or long-term monthly billing.

We recommend deploying private instances in the following scenarios:

  • maximize endpoint performance,
  • enable full context for long sequences,
  • ensure top-tier security for data processing in an isolated, dedicated environment,
  • use custom weights, such as fine-tuned models or LoRA adapters.

Recommended server configurations for hosting gemma-4-E4B-it

Prices:
Name GPU Price, hour TPS Max Concurrency
teslat4-1.16.16.160
131,072.0
1 $0.33 1.045 Launch
teslaa2-1.16.32.160
131,072.0
1 $0.38 1.045 Launch
teslaa10-1.16.32.160
131,072.0
1 $0.53 4.134 Launch
rtx2080ti-2.12.64.160
131,072.0
tensor
2 $0.69 2.289 Launch
rtx3090-1.16.24.160
131,072.0
1 $0.83 4.134 Launch
rtx3080-2.16.32.160
131,072.0
tensor
2 $0.97 1.517 Launch
rtx4090-1.16.32.160
131,072.0
1 $1.02 4.134 Launch
teslav100-1.12.64.160
131,072.0
1 $1.20 7.223 Launch
rtxa5000-2.16.64.160.nvlink
131,072.0
tensor
2 $1.23 12.328 Launch
rtx5090-1.16.64.160
131,072.0
1 $1.59 7.223 Launch
teslaa100-1.16.64.160
131,072.0
1 $2.37 25.756 Launch
h100-1.16.64.160
131,072.0
1 $3.83 25.756 Launch
h100nvl-1.16.96.160
131,072.0
1 $4.11 31.161 Launch
h200-1.16.128.160
131,072.0
1 $4.74 49.308 Launch
Prices:
Name GPU Price, hour TPS Max Concurrency
teslaa10-1.16.32.160
131,072.0
1 $0.53 3.061 Launch
teslat4-2.16.32.160
131,072.0
tensor
2 $0.54 5.078 Launch
teslaa2-2.16.32.160
131,072.0
tensor
2 $0.57 5.078 Launch
rtx2080ti-2.12.64.160
131,072.0
tensor
2 $0.69 1.217 Launch
rtx3090-1.16.24.160
131,072.0
1 $0.83 3.061 Launch
rtx4090-1.16.32.160
131,072.0
1 $1.02 3.061 Launch
teslav100-1.12.64.160
131,072.0
1 $1.20 6.150 Launch
rtxa5000-2.16.64.160.nvlink
131,072.0
tensor
2 $1.23 11.255 Launch
rtx3080-3.16.64.160
131,072.0
pipeline
3 $1.43 3.233 Launch
rtx5090-1.16.64.160
131,072.0
1 $1.59 6.150 Launch
rtx3080-4.16.64.160
131,072.0
tensor
4 $1.82 6.022 Launch
teslaa100-1.16.64.160
131,072.0
1 $2.37 24.683 Launch
h100-1.16.64.160
131,072.0
1 $3.83 24.683 Launch
h100nvl-1.16.96.160
131,072.0
1 $4.11 30.089 Launch
h200-1.16.128.160
131,072.0
1 $4.74 48.235 Launch
Prices:
Name GPU Price, hour TPS Max Concurrency
teslaa10-1.16.32.160
131,072.0
1 $0.53 1.804 Launch
teslat4-2.16.32.160
131,072.0
tensor
2 $0.54 3.821 Launch
teslaa2-2.16.32.160
131,072.0
tensor
2 $0.57 3.821 Launch
rtx3090-1.16.24.160
131,072.0
1 $0.83 1.804 Launch
rtx2080ti-3.12.24.120
131,072.0
pipeline
3 $0.84 3.134 Launch
rtx4090-1.16.32.160
131,072.0
1 $1.02 1.804 Launch
rtx2080ti-4.16.32.160
131,072.0
tensor
4 $1.12 6.309 Launch
teslav100-1.12.64.160
131,072.0
1 $1.20 4.893 Launch
rtxa5000-2.16.64.160.nvlink
131,072.0
tensor
2 $1.23 9.998 Launch
rtx3080-3.16.64.160
131,072.0
pipeline
3 $1.43 1.976 Launch
rtx5090-1.16.64.160
131,072.0
1 $1.59 4.893 Launch
rtx3080-4.16.64.160
131,072.0
tensor
4 $1.82 4.764 Launch
teslaa100-1.16.64.160
131,072.0
1 $2.37 23.426 Launch
h100-1.16.64.160
131,072.0
1 $3.83 23.426 Launch
h100nvl-1.16.96.160
131,072.0
1 $4.11 28.831 Launch
h200-1.16.128.160
131,072.0
1 $4.74 46.978 Launch

Related models

Need help?

Contact our dedicated neural networks support team at nn@immers.cloud or send your request to the sales department at sale@immers.cloud.