Browse all Google text models
Provider logo

Gemma 4 31B Thinking

google/gemma-4-31b-it:thinking
Provider logo

Gemma 4 31B Thinking

google/gemma-4-31b-it:thinking

Google's Gemma 4 31B instruction-tuned model with thinking explicitly enabled, exposing reasoning traces for complex multimodal and coding workflows. Requests containing video cost $0.14 per million input tokens and $0.40 per million output tokens.

Added Apr 2, 2026

Model weights

Context Window

262.1K

Max Output

131.1K

Avg output tokens (7d)

1.4K tokens

80%

Input Price (Auto)

$0.10/1M

Output Price (Auto)

$0.35/1M

Cache Read (Auto)

$0.050/1M

Capabilities

Benchmarks

Performance metrics and benchmarks

Sourced from Artificial Analysis.

Intelligence Index

15.4

Better than 60% of models compared

Coding Index

43.4

Better than 50% of models compared

Agentic Index

6.7

Better than 36% of models compared

Agentic work

AutomationBench-AA

Workflow automation with guardrail penalties

4.9%

Better than 32% of models compared

AA-Briefcase

Agentic knowledge work (Elo)

377 Elo

Better than 21% of models compared

GDPval-AA v2

Economically valuable tasks (Elo)

755 Elo

Better than 35% of models compared

Document reasoning

GDP.pdf

Professional PDF reasoning: all-pass rate

6.0%

Better than 34% of models compared

AA-LCR v1.1

Long context reasoning with updated grading

69.7%

Better than 66% of models compared

Reasoning

HLE

Humanity's Last Exam

23.6%

Better than 75% of models compared

IFBench

Instruction-following benchmark

75.6%

Better than 93% of models compared

CritPt

Research-level physics reasoning

1.4%

Coding

Terminal-Bench v4.0

Practical coding and terminal tasks

0.0%

Better than 17% of models compared

SciCode

Python programming for scientific computing

45.5%

Better than 44% of models compared

Knowledge

AA-Omniscience Accuracy

Proportion of correctly answered questions

20.0%

AA-Omniscience Hallucination Rate

Rate of incorrect answers among non-correct responses

85.0%

Legacy benchmarks

GPQA Diamond (legacy)

Graduate-level scientific reasoning

85.7%

Better than 79% of models compared

Terminal-Bench Hard (legacy)

Agentic coding and terminal use

36.4%

Better than 83% of models compared

T²-Bench Telecom (legacy)

Conversational AI agents in dual-control scenarios

59.9%

Better than 56% of models compared

AA-LCR (unversioned / legacy)

Long context reasoning evaluation

69.7%

Better than 66% of models compared

GDPval-AA (unversioned / legacy)

Economically valuable tasks

12.8%

Last updated Sep 13, 2026

Artificial Analysis

Providers

Choose explicit providers for this model. Auto routing remains available as the default option.

Loading provider options…

Compare Gemma 4 31B Thinking with similar models from the same provider or model family.