LLM Pricing
PricingLeaderboardsToolsProvidersReleasesGuides

© 2026 LLM Pricing

About
·Contact
·Privacy
·RSS
← Model list

Gemma 3 12B IT

google·google/gemma-3-12b-it·GA·Open weights·gemma series
Open multimodal Gemma instruction model for multilingual text generation and image understanding

Gemma 3 12B IT by google is offered by 12 providers on this page. Public prices are shown for 11 of them. Its reference price is $0.272 per 1M tokens. The lowest paid channel is NovitaAI at $0.05 / $0.10 per 1M, about 5.4× below the reference price. That channel is a third-party gateway, so confirm its availability and rate limits before depending on it. It also has 1 free ($0) channel; free tiers usually carry rate limits, and subscription-covered access bills $0 per token only after the subscription fee.

Artificial Analysis rates it 3.8 on the Intelligence Index, with 5.8 for coding and 0.1 for agentic tasks. Against its lowest blended price of $0.063 per 1M, that is roughly 61 index points per dollar, which is the value ratio the leaderboards rank on.

The context window is 131,072 tokens at the reference host, but hosts report different limits, from 128,000 to 131,072, so the usable window depends on the provider you pick. It supports tool use and structured output. Accepted input modalities are Text, Image, and PDF. The weights are open, so it can also be self-hosted or served through a gateway of your choice. Providers report the capability flags inconsistently, so verify a specific feature against the host you plan to use. Its training knowledge cuts off at 2024-08.

Specs & pricing

Input / output per 1M tokens
Reference price·NanoGPT
$0.27 / $0.27
Blended $0.27 · Cache read $0.14
Lowest paid·NovitaAIGateway
$0.05 / $0.10
Blended $0.063 · 5.4× spread
1 more $0 channels
Context
131,072
Output limit
16,384
Knowledge cutoff
2024-08
Released / updated
2025-03-12 / 2025-03-12
Capabilities
Reasoning✓ Tool use✓ Structured output✓ Temperature✓ Attachments
⚠ Providers report capability flags inconsistently
Modalities
TextImagePDF
Weights
Hugging Face

Quality & performanceArtificial Analysis · Intelligence Index v4.3

Intelligence3.8
Value61
Coding5.8
Agentic0.1
Output speed—
TTFT—
Value formulaIQ 3.8 ÷ min blended $0.063 = 61

Quality is independently evaluated by Artificial Analysis. Speed/latency are model-level medians.

Available at 12 providers11 with public prices · 1 free

ProviderTierInputOutputCache readCache writeContextOutput limitStatus
NvidiaCloudFree——131,07216,384
NovitaAIGateway$0.05$0.10——131,0728,192
HeliconeGateway$0.05$0.10——131,0728,192
Deep InfraCloud$0.05$0.15——131,072131,072
Kilo GatewayGateway$0.05$0.15——131,07216,384
OpenRouterGateway$0.05$0.15——131,07216,384
Hugging FaceCloud$0.05$0.15——131,072131,072
Eden AIGateway$0.05$0.15——131,072131,072
Amazon Bedrock
google.gemma-3-12b-it
Cloud$0.09$0.29——131,0728,192
Merge GatewayGateway$0.09$0.29——128,000 ⚠8,192
Neon
gemma-3-12b
Cloud$0.15$0.50——131,0728,192
NanoGPTGateway$0.27$0.27$0.14—131,07216,384

Sorted by blended price (input×0.75 + output×0.25) asc. The official channel always shows regardless of rank. Whether a gateway's low price is actually usable can't be verified.

Your usage cost

1Helicone$15.00
2NovitaAI$15.00
3Deep Infra$17.50
4Kilo Gateway$17.50
5OpenRouter$17.50
6Hugging Face$17.50

1 more channels offer $0 (Nvidia); free tiers usually have rate limits and no SLA, excluded from ranking.

The cheapest paid channel is the only channel.

Price historyone sample accumulated per data sync

Input listOutput listMin blended
$0.56$02026-08-052026-09-112026-08-05 · Input list $0.352026-08-06 · Input list $0.352026-08-07 · Input list $0.352026-08-08 · Input list $0.352026-08-09 · Input list $0.352026-08-10 · Input list $0.352026-08-11 · Input list $0.352026-08-12 · Input list $0.352026-08-13 · Input list $0.352026-08-14 · Input list $0.272026-09-03 · Input list $0.272026-09-05 · Input list $0.152026-09-11 · Input list $0.272026-08-05 · Output list $0.562026-08-06 · Output list $0.562026-08-07 · Output list $0.562026-08-08 · Output list $0.562026-08-09 · Output list $0.562026-08-10 · Output list $0.562026-08-11 · Output list $0.562026-08-12 · Output list $0.562026-08-13 · Output list $0.562026-08-14 · Output list $0.272026-09-03 · Output list $0.272026-09-05 · Output list $0.502026-09-11 · Output list $0.272026-08-05 · Min blended $0.272026-08-06 · Min blended $0.272026-08-07 · Min blended $0.272026-08-08 · Min blended $0.272026-08-09 · Min blended $0.272026-08-10 · Min blended $0.272026-08-11 · Min blended $0.272026-08-12 · Min blended $0.272026-08-13 · Min blended $0.272026-08-14 · Min blended $0.272026-09-03 · Min blended $0.0632026-09-05 · Min blended $0.0632026-09-11 · Min blended $0.063

Artificial Analysis evaluations14 items

GPQA Diamond34.9%
Humanity's Last Exam4.2%
MMLU-Pro59.5%
SciCode16.4%
LiveCodeBench13.7%
Terminal-Bench Hard0.8%
Terminal-Bench 2.10.0%
AIME 202518.3%
AIME22.0%
MATH-50085.3%
τ²-Bench Telecom10.8%
τ³-Bench Banking0.8%
AA-LCR8.3%
IFBench36.7%

Individual evaluations run by Artificial Analysis, on the same reasoning tier as the intelligence score above. Each benchmark has its own task set and harness, so rows are not comparable with one another. The Intelligence Index above draws on a different, newer set of evaluations.

Related models

Gemma 4 26B A4B Cybersecuritysame series$0.11 / $0.33Gemma 4 31B Split-Untiedsame series$0.10 / $0.30Gemma 4 12B Semancersame series$0.05 / $0.25Mistral Nemocheaper alternative$0.15 / $0.15Ministral 3Bcheaper alternative$0.04 / $0.04Command R7Bcheaper alternative$0.038 / $0.15Mistral Small 3.2cheaper alternative$0.10 / $0.30
Data partly from models.dev (MIT) · NanoGPT official docs ↗