Gemma 3 27B
Google · released Mar 1, 2025 · google/gemma-3-27b-it
- Type
- Open weightsGemma Terms of Use
- Params
- 27.4B
- Context
- 262K
about 197K words of context · download allowed, licence restricts use
Our take
Written Sep 2, 2026Gemma 3 is a 27.4-billion-parameter text-and-image model from Google with a 262,144-token request limit and measured scores across six Arena leaderboards. Its licence permits commercial use with restrictions, and it sits in the budget tier among tracked offers.
Pick this for general text-and-image workloads where a very large request limit matters and cost is a priority, or for throughput-sensitive tasks where you can trade price for speed. Skip it if you need fully unrestricted redistribution, if maths-heavy work dominates, or if you want the cheapest host to also be the fastest.
The case for it
- Six Arena variants measured with consistent mid-table scores: overall 1365.16, hard prompts 1365.84, coding 1357.95, creative writing 1348.21, instruction following 1343.35, maths 1322.10.
- 262,144-token request limit, unusually large for this parameter class.
- Lowest-cost tracked offer is well under half the price of the most expensive one.
The case against it
- Maths is its weakest measured variant, 43.7 points below its own hard-prompts score.
- The fastest host costs nearly three times the output price of the cheapest one.
- Gemma Terms of Use constrain redistribution and derivative use versus fully permissive licences.
How good is it?
An open text model from Google for everyday questions and writing, though coding is not its strong point.
- writing and completing codeArena Coding · 140th of 168
EverydayGeneral questions and everyday reasoning
Arena Text (overall)115th of 168 · 1365
CodingWriting and fixing code on its own
Arena Coding140th of 168 · 1358
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 Writing103rd of 168 · 1348
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. 3.6 GB spare means a 10% error in the size would not change the answer.
GeForce RTX 5090 · 32 GB
Room to spare. 11.6 GB spare means a 10% error in the size would not change the answer.
Apple M1 Pro (16-core GPU) · 32 GB
Room to spare. 4.8 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 2 hours ago — each listing carries its own date.
The only listing at 131K of context — the other 4 in the table below are not like-for-like. 3 cheaper rows there are outside that comparison: a different context length or a different quantisation.
- per 1M tokens
- $0.080 in / $0.45 out
- Context served
- 131K
- Throughput
- Not measured
| Provider | In / out per 1M tokens | Context | Throughput | Trains on prompts | Logs prompts | Zero retention |
|---|---|---|---|---|---|---|
| DeepInfrafp8Direct and through OpenRouter | $0.080 / $0.16checked 2 hours ago | 131K16K max reply through OpenRouter | 26 tok/sthrough OpenRouter | DirectUnknownThrough OpenRouterNo | DirectUnknownThrough OpenRouterNo | DirectUnknownThrough OpenRouterConfirmed |
| Novita AIbf16Direct and through OpenRouter | $0.12 / $0.20checked 2 hours ago | 98K16K max reply through OpenRouter | 28 tok/sthrough OpenRouter | DirectUnknownThrough OpenRouterNo | DirectUnknownThrough OpenRouterNo | DirectUnknownThrough OpenRouterConfirmed |
| Nebius AI Studiofp8Through OpenRouter | $0.10 / $0.30checked 2 hours ago | 110K99K max reply | 21 tok/s | No | No | Confirmed |
| Parasailfp8Through OpenRouter | $0.080 / $0.45checked 2 hours ago | 131K118K max reply | 36 tok/s | No | No | Confirmed |
| OpenRouterOpenRouter's own listing | $0.080 / $0.45checked 2 hours ago | 131K | not measured | Unknown | Unknown | Unknown |
DeepInfra
Novita AI
Nebius AI StudioAcross the 5 listings we hold: 4 say they do not train on prompts (2 of them only through OpenRouter), 0 say they do and 1 does not say. 4 appear in the zero-retention registry we check (2 of them only through OpenRouter); 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 |
|---|---|---|---|
| DeepInfrafp8Direct and through OpenRouter | ✓ | ✓ | ✓ |
| Novita AIbf16Direct and through OpenRouter | ✗ | ✗ | ✗ |
| Nebius AI Studiofp8Through OpenRouter | ✗ | ✓ | ✓ |
| Parasailfp8Through OpenRouter | ✗ | ✓ | ✓ |
| OpenRouterOpenRouter's own listing | ✓ | ✓ | ✓ |
Tool calling: 2 of 5 listings say yes, 3 say no. JSON output: 4 of 5 listings say yes, 1 says no. Strict schema: 4 of 5 listings say yes, 1 says no.
Models people weigh against Gemma 3 27B
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 5 listings does not say whether it trains on prompts, and 2 answer only through OpenRouter, not for their own listing.
- 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-3-27b-it
- Takes in, gives back
- Text and images in, text out
- Catalogue slug
- google-gemma-3-27b