Gemma 3n 4B
Google · released Jun 3, 2025 · google/gemma-3n-E4B-it
- Type
- Open weightsGemma Terms of Use
- Params
- 7.8B
- Context
- 33K
about 25K words of context · download allowed, licence restricts use
Our take
Written Sep 2, 2026Gemma 3n is a small downloadable text model from Google with measured performance across six Arena leaderboards. Its pricing is identical across both tracked hosts, making it a straightforward entry-level pick for lightweight inference.
Pick this for low-cost hosted inference where a small, Google-backed model with broad evaluation coverage is enough. Use it on Together.ai if 35 tokens per second meets your latency needs. Skip it if you need a permissive licence, strong maths performance, or multimodal input.
The case for it
- Six measured Arena dimensions for its size class: overall, coding, hard prompts, creative writing, instruction following and maths.
- Identical pricing across both tracked providers, so there is no price arbitrage to hunt.
The case against it
- Maths is the weakest measured dimension, with a 57.7-point gap below its own overall score.
- Throughput on Together.ai is modest at 35 tokens per second; speed on the other provider is unverified in our data.
- Gemma Terms of Use carry commercial restrictions, not a permissive open licence.
How good is it?
An open small text model for light chat and simple tasks, though it trails most models on everyday questions, writing and code.
- getting answers to everyday questionsArena Text (overall) · 141st of 168
- drafts, rewrites and editingArena Creative Writing · 130th of 168
- writing and completing codeArena Coding · 154th of 168
EverydayGeneral questions and everyday reasoning
Arena Text (overall)141st of 168 · 1317
CodingWriting and fixing code on its own
Arena Coding154th of 168 · 1307
Arena Coding is the only board that has scored it for this.
AgenticPlanning, calling tools, staying on task
Not yet scored on Arena Agent.
WritingDrafting and rewriting prose
Arena Creative Writing130th of 168 · 1299
Arena Creative Writing is the only board that has scored it for this.
These tests check whether a model follows instructions — a precondition for all the work above, but not a measure of how well that work is done.
Every published score for this model6 scoresEvery figure we hold, from 6 boards, with who ran it and a link to the source — including the boards no rating above is built on.
Can you run it yourself?
Comfortable fit
GeForce RTX 4090 · 24 GB
Room to spare. 16.4 GB spare means a 10% error in the size would not change the answer.
GeForce RTX 5090 · 32 GB
Room to spare. 24.4 GB spare means a 10% error in the size would not change the answer.
Apple M1 (8-core GPU) · 16 GB
Room to spare. 5.6 GB spare means a 10% error in the size would not change the answer.
Memory use by level
Against a 24 GB card.
What is quantisation? →This model on every device we track71 devicesThe Q4 build most people download, on each device: what the weights come to, how much context the memory leaves, and whether it runs. Smallest device that runs it first.
Check against your own machine → · Where to rent it hosted →
Or rent it from someone else
Prices checked 37 days ago — each listing carries its own date.
- per 1M tokens
- $0.060 in / $0.12 out
- Context served
- 33K
- Throughput
- Not measured
| Provider | In / out per 1M tokens | Context | Throughput | Trains on prompts | Logs prompts | Zero retention |
|---|---|---|---|---|---|---|
| OpenRouterOpenRouter's own listing | $0.060 / $0.12checked 37 days ago | 33K | not measured | Unknown | Unknown | Unknown |
| Together AIThrough OpenRouter | $0.060 / $0.12checked 37 days ago | 33K | not measured | No | No | Unknown |
OpenRouterAcross the 2 listings we hold: 1 says it does not train on prompts, 0 say they do and 1 does not say. 0 appear in the zero-retention registry we check; the rest are unknown to us.
What each host's API supports
From the parameter list each endpoint publishes. Streaming is omitted: nothing we hold reports it, for any model.
| Provider | Tool calling | JSON output | Strict schema |
|---|---|---|---|
| OpenRouterOpenRouter's own listing | ✗ | ✓ | ✓ |
| Together AIThrough OpenRouter | ✗ | ✓ | ✓ |
Tool calling: 0 of 2 listings say yes, 2 say no. JSON output: 2 of 2 listings say yes. Strict schema: 2 of 2 listings say yes.
Models people weigh against Gemma 3n 4B
When we formed this view
Recent changes
What moved
first indexed by our pipelineEach date is the day we first saw the change, or the day the maker announced it.
What we do not know about this model yet
- We hold no measured file for it, so all 3 sizes on this page are calculated from the parameter count.
- Nothing we hold says whether an endpoint streams, so we do not show it either way.
- 1 of 2 listings does not say whether it trains on prompts.
- We hold no cached-input rate for any of its listings.
- We hold no batch or off-peak rate for any of its listings.
- We hold a decode speed for it, but no prompt-processing (prefill) figure, so how long the input side of a job takes is unknown to us.
Licence and identifiers
What the licence allowsGemma Terms of Use, what it allows commercially, and the identifiers you need to pull this model — its Hugging Face repo, our slug and a machine-readable card.
Licence
Gemma Terms of Use
Commercial use allowed, but Google's prohibited-use policy applies and can be updated over time — terms are less static than Apache/MIT.
Identifiers
- Hugging Face
- google/gemma-3n-E4B-it
- Takes in, gives back
- Text in, text out
- Catalogue slug
- google-gemma-3n-4b