Google: Gemma 3 12B

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,…

google/gemma-3-12b-it

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)

Specifications & capabilities

Input
text, image
Output
text
Context window · tokens
131072
Knowledge cutoff
2024-08-31
Published model ID
gemma3@gemma-3-12b-it
Capabilities
multimodal, structuredOutputs, toolCalling, visionInput
Description source ↗

Evaluations

21–40 / 173

MMBench

73.6%

source id
mmBench
leaderboard name
future_prediction
dimension
vision
metric
future_prediction
rank
50
aggregation scope
status
published
direction
higherIsBetter
conditions
suite
MMBench_TEST_CN_Data
verified
Yes
metric_kind
future_prediction
rank_method
competition rank among reported evaluations in identical conditions, including unlinked observations
source_file
mmbench-data.json
source_revision
acb08c72ac038d24cf7785254678efe725354fe91091bfa10a667832ad964b24
metric_population
163
benchmark_evaluation
true
source model id
Gemma3-12B
Original result ↗

OpenRouter Benchmarks

48%

source id
openRouterBenchmarks
leaderboard name
tau2-bench Verified Airline · Accuracy · DeepInfra
dimension
agentic
metric
tau2-bench Verified Airline · Accuracy
rank
50
sample size
50
aggregation scope
status
published
harness
url
https://github.com/OpenRouterTeam/benchmark-harness
name
OpenRouter benchmark harness
direction
higherIsBetter
run_count
1
conditions
window
newest 90 days; successful visible nonzero runs
dataset
benchmark-leaderboard
benchmark
id
tau_bench_verified_airline
name
tau2-bench Verified Airline
record_kind
aggregate
provider_name
DeepInfra
attribution
Source: OpenRouter (openrouter.ai), licensed under CC BY 4.0: reuse and republish with attribution to OpenRouter.
last_run_at
2026-10-08T22:08:03.352Z
license_url
https://creativecommons.org/licenses/by/4.0/
methodology
Rows aggregate the benchmark-runs dataset exactly as the leaderboard pages do, over each benchmark's non-zero-score runs from the newest 90 days: accuracy is the task-weighted mean of per-run accuracy, accuracy_std_dev is the population standard deviation of per-run accuracy (null with a single run), and the per-task averages divide summed cost, generation time, and output tokens by summed tasks. Rows are ordered by accuracy descending, then model name.
metric_kind
tau2-bench Verified Airline · Accuracy
rank_method
competition rank among reported evaluations in identical conditions, including unlinked observations
sample_size
50
source_file
benchmark-leaderboard.json
source_page
https://openrouter.ai/benchmarks/tau2-bench-airline
measurements
accuracy
0.48
cost_per_task_usd
0.011001320000000002
output_tokens_per_task
3762.88
latency_seconds_per_task
80.20564
published_at
2026-11-16T01:46:02.653Z
snapshot_date
2026-11-16
source_revision
7ccaf17664252a3053dcb7df8ad0a46e72954a3b4524ddc1c8a7ef8a41b15568
transformations
TensorTrace preserves source results, converts latency to seconds and derives ranks within matching conditions. Individual runs have no derived rank.
metric_population
63
benchmark_evaluation
true
openrouter_permaslug
google/gemma-3-12b-it
source model id
google/gemma-3-12b-it
Original result ↗

MMBench

84.7%

source id
mmBench
leaderboard name
celebrity_recognition
dimension
vision
metric
celebrity_recognition
rank
51
aggregation scope
status
published
direction
higherIsBetter
conditions
suite
MMBench_TEST_CN_Data
verified
Yes
metric_kind
celebrity_recognition
rank_method
competition rank among reported evaluations in identical conditions, including unlinked observations
source_file
mmbench-data.json
source_revision
acb08c72ac038d24cf7785254678efe725354fe91091bfa10a667832ad964b24
metric_population
163
benchmark_evaluation
true
source model id
Gemma3-12B
Original result ↗

MMBench

87.7%

source id
mmBench
leaderboard name
function_reasoning
dimension
vision
metric
function_reasoning
rank
58
aggregation scope
status
published
direction
higherIsBetter
conditions
suite
MMBench_TEST_EN_Data
verified
Yes
metric_kind
function_reasoning
rank_method
competition rank among reported evaluations in identical conditions, including unlinked observations
source_file
mmbench-data.json
source_revision
acb08c72ac038d24cf7785254678efe725354fe91091bfa10a667832ad964b24
metric_population
162
benchmark_evaluation
true
source model id
Gemma3-12B
Original result ↗

MMBench

86.8%

source id
mmBench
leaderboard name
function_reasoning
dimension
vision
metric
function_reasoning
rank
59
aggregation scope
status
published
direction
higherIsBetter
conditions
suite
MMBench_TEST_CN_Data
verified
Yes
metric_kind
function_reasoning
rank_method
competition rank among reported evaluations in identical conditions, including unlinked observations
source_file
mmbench-data.json
source_revision
acb08c72ac038d24cf7785254678efe725354fe91091bfa10a667832ad964b24
metric_population
163
benchmark_evaluation
true
source model id
Gemma3-12B
Original result ↗

MMBench

70.5%

source id
mmBench
leaderboard name
LR
dimension
vision
metric
LR
rank
60
aggregation scope
status
published
direction
higherIsBetter
conditions
suite
MMBench_TEST_EN_Data
verified
Yes
metric_kind
LR
rank_method
competition rank among reported evaluations in identical conditions, including unlinked observations
source_file
mmbench-data.json
source_revision
acb08c72ac038d24cf7785254678efe725354fe91091bfa10a667832ad964b24
metric_population
235
benchmark_evaluation
true
source model id
Gemma3-12B
Original result ↗

MMBench

67.3%

source id
mmBench
leaderboard name
structuralized_imagetext_understanding
dimension
vision
metric
structuralized_imagetext_understanding
rank
60
aggregation scope
status
published
direction
higherIsBetter
conditions
suite
MMBench_TEST_EN_Data
verified
Yes
metric_kind
structuralized_imagetext_understanding
rank_method
competition rank among reported evaluations in identical conditions, including unlinked observations
source_file
mmbench-data.json
source_revision
acb08c72ac038d24cf7785254678efe725354fe91091bfa10a667832ad964b24
metric_population
162
benchmark_evaluation
true
source model id
Gemma3-12B
Original result ↗

MMBench

97.8%

source id
mmBench
leaderboard name
image_topic
dimension
vision
metric
image_topic
rank
61
aggregation scope
status
published
direction
higherIsBetter
conditions
suite
MMBench_TEST_CN_V11_Data
verified
Yes
metric_kind
image_topic
rank_method
competition rank among reported evaluations in identical conditions, including unlinked observations
source_file
mmbench-data.json
source_revision
acb08c72ac038d24cf7785254678efe725354fe91091bfa10a667832ad964b24
metric_population
295
benchmark_evaluation
true
source model id
Gemma3-12B
Original result ↗

MMBench

79.3%

source id
mmBench
leaderboard name
image_style
dimension
vision
metric
image_style
rank
61
aggregation scope
status
published
direction
higherIsBetter
conditions
suite
MMBench_TEST_CN_V11_Data
verified
Yes
metric_kind
image_style
rank_method
competition rank among reported evaluations in identical conditions, including unlinked observations
source_file
mmbench-data.json
source_revision
acb08c72ac038d24cf7785254678efe725354fe91091bfa10a667832ad964b24
metric_population
295
benchmark_evaluation
true
source model id
Gemma3-12B
Original result ↗

MMBench

76.3%

source id
mmBench
leaderboard name
RR
dimension
vision
metric
RR
rank
63
aggregation scope
status
published
direction
higherIsBetter
conditions
suite
MMBench_TEST_CN_Data
verified
Yes
metric_kind
RR
rank_method
competition rank among reported evaluations in identical conditions, including unlinked observations
source_file
mmbench-data.json
source_revision
acb08c72ac038d24cf7785254678efe725354fe91091bfa10a667832ad964b24
metric_population
236
benchmark_evaluation
true
source model id
Gemma3-12B
Original result ↗

MMBench

93.2%

source id
mmBench
leaderboard name
action_recognition
dimension
vision
metric
action_recognition
rank
66
aggregation scope
status
published
direction
higherIsBetter
conditions
suite
MMBench_TEST_EN_Data
verified
Yes
metric_kind
action_recognition
rank_method
competition rank among reported evaluations in identical conditions, including unlinked observations
source_file
mmbench-data.json
source_revision
acb08c72ac038d24cf7785254678efe725354fe91091bfa10a667832ad964b24
metric_population
162
benchmark_evaluation
true
source model id
Gemma3-12B
Original result ↗

MMBench

89.1%

source id
mmBench
leaderboard name
nature_relation
dimension
vision
metric
nature_relation
rank
67
aggregation scope
status
published
direction
higherIsBetter
conditions
suite
MMBench_TEST_CN_V11_Data
verified
Yes
metric_kind
nature_relation
rank_method
competition rank among reported evaluations in identical conditions, including unlinked observations
source_file
mmbench-data.json
source_revision
acb08c72ac038d24cf7785254678efe725354fe91091bfa10a667832ad964b24
metric_population
295
benchmark_evaluation
true
source model id
Gemma3-12B
Original result ↗

MMBench

81.5%

source id
mmBench
leaderboard name
nature_relation
dimension
vision
metric
nature_relation
rank
67
aggregation scope
status
published
direction
higherIsBetter
conditions
suite
MMBench_TEST_EN_Data
verified
Yes
metric_kind
nature_relation
rank_method
competition rank among reported evaluations in identical conditions, including unlinked observations
source_file
mmbench-data.json
source_revision
acb08c72ac038d24cf7785254678efe725354fe91091bfa10a667832ad964b24
metric_population
162
benchmark_evaluation
true
source model id
Gemma3-12B
Original result ↗

OpenRouter Benchmarks

40.572%

source id
openRouterBenchmarks
leaderboard name
GPQA Diamond · Accuracy · DeepInfra
dimension
knowledge
metric
GPQA Diamond · Accuracy
rank
68
sample size
198
aggregation scope
status
published
harness
url
https://github.com/OpenRouterTeam/benchmark-harness
name
OpenRouter benchmark harness
direction
higherIsBetter
run_count
1
conditions
window
newest 90 days; successful visible nonzero runs
dataset
benchmark-leaderboard
benchmark
id
gpqa_diamond
name
GPQA Diamond
record_kind
aggregate
provider_name
DeepInfra
attribution
Source: OpenRouter (openrouter.ai), licensed under CC BY 4.0: reuse and republish with attribution to OpenRouter.
last_run_at
2026-10-08T22:01:16.852Z
license_url
https://creativecommons.org/licenses/by/4.0/
methodology
Rows aggregate the benchmark-runs dataset exactly as the leaderboard pages do, over each benchmark's non-zero-score runs from the newest 90 days: accuracy is the task-weighted mean of per-run accuracy, accuracy_std_dev is the population standard deviation of per-run accuracy (null with a single run), and the per-task averages divide summed cost, generation time, and output tokens by summed tasks. Rows are ordered by accuracy descending, then model name.
metric_kind
GPQA Diamond · Accuracy
rank_method
competition rank among reported evaluations in identical conditions, including unlinked observations
sample_size
198
source_file
benchmark-leaderboard.json
source_page
https://openrouter.ai/benchmarks/gpqa-diamond
measurements
accuracy
0.405724
cost_per_task_usd
0.0003259848484848485
output_tokens_per_task
1917.1515151515152
latency_seconds_per_task
29.056348484848485
published_at
2026-11-16T01:46:02.653Z
snapshot_date
2026-11-16
source_revision
7ccaf17664252a3053dcb7df8ad0a46e72954a3b4524ddc1c8a7ef8a41b15568
transformations
TensorTrace preserves source results, converts latency to seconds and derives ranks within matching conditions. Individual runs have no derived rank.
metric_population
71
benchmark_evaluation
true
openrouter_permaslug
google/gemma-3-12b-it
source model id
google/gemma-3-12b-it
Original result ↗

MMBench

97.4%

source id
mmBench
leaderboard name
identity_reasoning
dimension
vision
metric
identity_reasoning
rank
69
aggregation scope
status
published
direction
higherIsBetter
conditions
suite
MMBench_TEST_CN_V11_Data
verified
Yes
metric_kind
identity_reasoning
rank_method
competition rank among reported evaluations in identical conditions, including unlinked observations
source_file
mmbench-data.json
source_revision
acb08c72ac038d24cf7785254678efe725354fe91091bfa10a667832ad964b24
metric_population
295
benchmark_evaluation
true
source model id
Gemma3-12B
Original result ↗

MMBench

77.6%

source id
mmBench
leaderboard name
image_style
dimension
vision
metric
image_style
rank
71
aggregation scope
status
published
direction
higherIsBetter
conditions
suite
MMBench_TEST_EN_Data
verified
Yes
metric_kind
image_style
rank_method
competition rank among reported evaluations in identical conditions, including unlinked observations
source_file
mmbench-data.json
source_revision
acb08c72ac038d24cf7785254678efe725354fe91091bfa10a667832ad964b24
metric_population
162
benchmark_evaluation
true
source model id
Gemma3-12B
Original result ↗

MMBench

46.9%

source id
mmBench
leaderboard name
spatial_relationship
dimension
vision
metric
spatial_relationship
rank
71
aggregation scope
status
published
direction
higherIsBetter
conditions
suite
MMBench_TEST_EN_Data
verified
Yes
metric_kind
spatial_relationship
rank_method
competition rank among reported evaluations in identical conditions, including unlinked observations
source_file
mmbench-data.json
source_revision
acb08c72ac038d24cf7785254678efe725354fe91091bfa10a667832ad964b24
metric_population
162
benchmark_evaluation
true
source model id
Gemma3-12B
Original result ↗

MMBench

78.7%

source id
mmBench
leaderboard name
future_prediction
dimension
vision
metric
future_prediction
rank
73
aggregation scope
status
published
direction
higherIsBetter
conditions
suite
MMBench_TEST_EN_V11_Data
verified
Yes
metric_kind
future_prediction
rank_method
competition rank among reported evaluations in identical conditions, including unlinked observations
source_file
mmbench-data.json
source_revision
acb08c72ac038d24cf7785254678efe725354fe91091bfa10a667832ad964b24
metric_population
295
benchmark_evaluation
true
source model id
Gemma3-12B
Original result ↗

MMBench

63%

source id
mmBench
leaderboard name
physical_property_reasoning
dimension
vision
metric
physical_property_reasoning
rank
76
aggregation scope
status
published
direction
higherIsBetter
conditions
suite
MMBench_TEST_CN_Data
verified
Yes
metric_kind
physical_property_reasoning
rank_method
competition rank among reported evaluations in identical conditions, including unlinked observations
source_file
mmbench-data.json
source_revision
acb08c72ac038d24cf7785254678efe725354fe91091bfa10a667832ad964b24
metric_population
163
benchmark_evaluation
true
source model id
Gemma3-12B
Original result ↗

MMBench

92.2%

source id
mmBench
leaderboard name
image_scene
dimension
vision
metric
image_scene
rank
77
aggregation scope
status
published
direction
higherIsBetter
conditions
suite
MMBench_TEST_CN_Data
verified
Yes
metric_kind
image_scene
rank_method
competition rank among reported evaluations in identical conditions, including unlinked observations
source_file
mmbench-data.json
source_revision
acb08c72ac038d24cf7785254678efe725354fe91091bfa10a667832ad964b24
metric_population
163
benchmark_evaluation
true
source model id
Gemma3-12B
Original result ↗

Published prices

1–6 / 6

gemma-3-12b-it ↗

$0.04

Inference provider
cheaperInference
Platform
Cheaper Inference
Unit
input_tokens_per_million
Currency
USD
Billing conditions
zdr
false
protocol
chat_completions
service_tier
standard
Billing label
Input tokens (per 1M)

gemma-3-12b-it ↗

$0.13

Inference provider
cheaperInference
Platform
Cheaper Inference
Unit
output_tokens_per_million
Currency
USD
Billing conditions
zdr
false
protocol
chat_completions
service_tier
standard
Billing label
Output tokens (per 1M)

google/gemma-3-12b-it ↗

$0.05

Inference provider
deepInfra
Platform
DeepInfra
Unit
input_tokens_per_million
Currency
USD
Billing conditions

google/gemma-3-12b-it ↗

$0.15

Inference provider
deepInfra
Platform
DeepInfra
Unit
output_tokens_per_million
Currency
USD
Billing conditions

google/gemma-3-12b-it ↗

$0.05

Inference provider
openRouter
Platform
OpenRouter
Unit
input_tokens_per_million
Currency
USD
Billing conditions

google/gemma-3-12b-it ↗

$0.15

Inference provider
openRouter
Platform
OpenRouter
Unit
output_tokens_per_million
Currency
USD
Billing conditions

Provider specifications & access

1–5 / 5

gemma-3-12b-it ↗

source model id
gemma-3-12b-it
hosting
gateway
platforms
  • Cheaper Inference
context length
128000
max output tokens
8192
architecture
input_modalities
  • text
  • image
output_modalities
  • text
capabilities
  • structuredOutputs
  • toolCalling
  • visionInput

Gemma 3 12B IT ↗

source model id
google.gemma-3-12b-it
hosting
bedrock-managed
platforms
  • Amazon Bedrock
architecture
input_modalities
  • text
  • image
output_modalities
  • text
capabilities

    google/gemma-3-12b-it ↗

    source model id
    google/gemma-3-12b-it
    hosting
    nvidia-inference-listed
    platforms
    • NVIDIA inference catalog
    architecture
    input_modalities
      output_modalities
        capabilities
          availability
          Listed by inference API; runtime availability unverified

          gemma-3-12b-it ↗

          source model id
          publishers/google/models/gemma3@gemma-3-12b-it
          hosting
          vertex-self-deployed
          platforms
          • Google Cloud Vertex AI
          architecture
          input_modalities
          • image
          • text
          output_modalities
          • text
          capabilities
          • multimodal
          • visionInput

          google/gemma-3-12b-it ↗

          source model id
          publishers/hf-google/models/gemma-3-12b-it@001
          hosting
          vertex-self-deployed
          platforms
          • Google Cloud Vertex AI
          architecture
          input_modalities
          • image
          • text
          output_modalities
          • text
          capabilities
          • multimodal
          • visionInput

          Safety assessments

          1–1 / 1

          Foundation Model Transparency Index 2025 ↗

          41 / 100

          source id
          stanfordFMTI
          canonical model id
          google/gemma-3-12b-it
          taxonomy
          transparency
          hazard category
          transparency
          hazard name
          Organization transparency
          policy area
          Google
          assessment type
          organization-level disclosure
          metric
          disclosure_score
          result label
          41 / 100
          sample size
          100
          updated at
          2026-11-09T18:08:05.227Z

          Source specification observations

          1–3 / 3

          openRouter ↗

          field
          context_length
          value
          131072
          normalized value
          131072
          scope
          source-model
          source id
          openRouter
          confidence
          1

          openRouter ↗

          field
          knowledge_cutoff
          reported field
          knowledge_cutoff
          value
          2024-08-31
          normalized value
          2024-08-31
          precision
          day
          scope
          source-model
          source id
          openRouter
          confidence
          1

          openRouter ↗

          field
          max_output_tokens
          value
          16384
          normalized value
          16384
          scope
          source-model
          source id
          openRouter
          confidence
          1

          Original sources

          1–20 / 21

          storage.googleapis.com · Model reference ↗

          Hugging Face · google/gemma-3-12b-it ↗

          DeepInfra · Model page ↗

          Google Cloud · Model page ↗

          Google Cloud · Model page ↗

          OpenRouter · Model page ↗

          api.cheaperinference.com · Source snapshot ↗

          Amazon Web Services · Model page ↗

          NVIDIA · Catalog ↗

          Amazon Bedrock · Model documentation ↗

          Amazon Web Services · Catalog ↗

          DeepInfra · Catalog ↗

          Epoch AI · Catalog ↗

          GitHub · Catalog ↗

          GitHub · Catalog ↗

          Google Cloud · Catalog ↗

          Hugging Face · Catalog ↗

          Hugging Face · Catalog ↗

          LMArena · Catalog ↗

          OpenRouter · Catalog ↗

          Source links

          1–15 / 15

          Explore more models

          Models

          Loading
          EVIDENCE / COMPARISON

          Benchmarks

          Explore every benchmark

          Loading

          These charts contain model-linked results. The charts above also include unlinked source profiles. Compare scores within each evaluation. Model quality, usage, transparency and system performance measure different things.

          Install TensorTrace