Google Released March 13, 2025 Synced Apr 19, 2026

Google: Gemma 3 12B (free)

Gemma 3 introduces multimodality, supporting vision-language input and text outputs. It handles context windows up to 128k tokens, understands over 140 languages, and offers improved math, reasoning, and chat capabilities,...

Vision

Why it stands out

Multimodal input (vision) extends it beyond text-only workflows.

$0/M input makes it practical for always-on agents, batch processing, or high-volume classification.

What to watch

No tool-use capability is currently tracked, which limits its fit for agentic or function-calling patterns.

33K context window is shorter than the longest-context frontier models available today.

No benchmark score currently tracked — evaluate using task-specific testing alongside pricing and capability data.

Release timeline

Tracked events for Google: Gemma 3 12B (free).

Back to model tracker

release

Google: Gemma 3 12B (free) entered the tracked catalog

March 13, 2025

Gemma 3 introduces multimodality, supporting vision-language input and text outputs. It handles context windows up to 128k tokens, understands over 140 languages, and offers improved math, reasoning, and chat capabilities, including structured outputs and function calling. Gemma 3 12B is the second largest in the family of Gemma 3 models after [Gemma 3 27B](google/gemma-3-27b-it)

Nearby alternatives

Other Google models worth checking.

Need a recommendation instead?

Google: Gemma 4 26B A4B

Gemma 4 26B A4B IT is an instruction-tuned Mixture-of-Experts (MoE) model from Google DeepMind. Despite 25.2B total parameters, only 3.8B ac

Context 262,144

Google: Gemma 4 26B A4B (free)

Gemma 4 26B A4B IT is an instruction-tuned Mixture-of-Experts (MoE) model from Google DeepMind. Despite 25.2B total parameters, only 3.8B ac

Context 262,144

Google: Gemma 4 31B

Gemma 4 31B Instruct is Google DeepMind's 30.7B dense multimodal model supporting text and image input with text output. Features a 256K tok

Context 262,144

Google: Gemma 4 31B (free)

Gemma 4 31B Instruct is Google DeepMind's 30.7B dense multimodal model supporting text and image input with text output. Features a 256K tok

Context 262,144

Recent changes

LaunchMar 13

Google launched Google: Gemma 3 12B (free)

Compare

See how Google: Gemma 3 12B (free) stacks up.

All comparisons

vs DeepSeek: DeepSeek V3 0324

Side-by-side pricing, context, and capabilities

vs DeepSeek: DeepSeek V3.1

Side-by-side pricing, context, and capabilities

vs DeepSeek: DeepSeek V3

Side-by-side pricing, context, and capabilities

vs DeepSeek: R1 Distill Llama 70B

Side-by-side pricing, context, and capabilities

vs DeepSeek: R1 Distill Qwen 32B

Side-by-side pricing, context, and capabilities

vs DeepSeek: DeepSeek V3.1 Terminus (exacto)

Side-by-side pricing, context, and capabilities