← Back to all tools

Gemma

Google's open-weights model family

APIFree tier
Visit site

Overall score

2.7/ 5
SME Fit2/5pricing pattern unclear + free tier · technical setup
JTBD4/5solid named JTBD
Integration4/5API + 7 integrations
Trust3/5mature, founded 2024
Quality1/5no public rating
Compliance1/5customer-choice residency

About

Gemma is Google's family of open-weights models, distilled from the same research as Gemini. Sized for efficient on-device and small-server inference. Strong reputation for quality at small parameter counts.

Best for: Builders who need a small open-weights model with strong quality at the 2B-27B range, on-device deployments, and a clean licence.

Pricing

TierMonthlyAnnual /moBillingNotes
Open Weightsn/aFreeflatAll Gemma weights;Apache 2.0 licence (recent variants);Run on-device, self-hosted, or any inference provider · Inference costs are paid to whichever provider hosts the model, or your own hardware.

Key features

  • Multiple sizes (2B to 27B+)
  • Apache 2.0 licence (newer variants)
  • Strong quality-to-size ratio
  • CodeGemma + RecurrentGemma specialised variants
  • Available on every major inference provider

Integrations

Hugging FaceTogether AIGroqReplicateOllamaVertex AIGoogle AI Studio

Trust & compliance

Stage range
n/a
Founded
2024
Status
active
SOC 2
n/a
GDPR
n/a
Data residency
customer_choice
External rating
n/a
Last verified
May 2026

Reviews

Be the first to share your experience.

Pairs well with