Gemma 3 27B IT
General purposeMultimodal
Aside from the points mentioned in our post, the Gemma release highlights some tricks used by labs: Knowledge Distillation using a big teacher model, a (5:1) local / global attention layer ratio. The latter is a configuration outlined by Noam Shazeer during his time at Character.AI. Apart from the instruction models, Google also releases the pre-trained base model.
Specs
Params27B
LicenseGemma
Capability · Artificial Analysis
AA Index
7.4
Months Behind Frontier
16.2 mo
gpt-4-turbo
06 Nov 2023
Adoption · Hugging Face
RAM score
—
Hugging Face Downloads
853.1K
last 30d
16.4M
all time
HF Likes
2K
Relative Adoption Metric contextualizes downloads against the model's size bucket.
Inference · OpenRouter
Tokens/Day
—
below top 50 this week
Peak Tokens/Day
9.4B
28 Jun 2026
Peak Rank
#19
30 Mar 2025
OpenRouter publishes daily token totals for its ~50 most-served models. Days missing from the chart mean Gemma 3 27B IT fell below that cutoff — not zero usage. OpenRouter logs usage separately per dated model version and per variant (like ":free"); the hub sums them into one daily total per model.
Related Models
