Gemma 4 31B
Google · 2 thinking configurations
Intelligence19.0
Input / 1M$0
Output / 1M$0
Context256K
Tariff: Google · Source: artificial-analysis · USD per million tokens
Thinking configuration
What does more thinking buy you?
Compare this model’s measured thinking configurations.
Scroll the chart ↔ · tap a point for details
Efficient frontier: best measured score at each cost2 measured configurations · 0 without both values
Selected: thinking thinking · Click a point to switch configuration
Gemma 4 31B Instruct is Google DeepMind's 30.7B dense multimodal model supporting text and image input with text output. Features a 256K token context window, configurable thinking/reasoning mode, native function...
Intelligence IndexArtificial Analysis composite index, not a percentage. Different versions of the index are not directly comparable.
19.0 pointsGPQA DiamondGraduate-level scientific questions. Measures specialist reasoning rather than web-search quality.
85.7 %Humanity’s Last ExamDifficult questions across many disciplines, compared under the same evaluation setup.
23.6 %SciCodeCode generation for scientific research problems.
45.5 %Terminal-Bench 2.1Agent tasks in a terminal environment. Results also depend on the evaluation harness.
43.4 %τ²-BenchTool use and interaction in assistance scenarios. Does not replace an evaluation of your integration.
59.9 %IFBenchVerifiable instruction following. A proxy for precision rather than a benchmark of creative style.
75.6 %MMMU-ProMultimodal understanding of academic problems that include images.
73.4 %Omniscience accuracyFactual accuracy in Artificial Analysis's Omniscience evaluation.
20.0 %Specifications from the catalog
Inputimage, text, video
Outputtext
Tool callingYes
Structured outputYes
Speed35.5 tok/s
First token1.00 s
Maximum output16.4K
Catalog listing date2026-04-02