Gemma 4 31B
Google • LLM#289 of 668 in intelligence
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...
Specifications
Context Window
262K tokens
Input Price/1M
—
#1 of 806
Output Price/1M
—
Parameters
—
Speed
35 tok/s
#389 of 651
Latency (TTFT)
1.1s
Max Output
16K tokens
Benchmarks
Gemma 4 31B results on the main AI model evaluation benchmarks. Higher scores indicate better performance.
Agentic
| Benchmark | Score | Maximum | Methodology |
|---|---|---|---|
| Terminal-Bench Hard | 36.0 | 100.0 | — |
| AA Agentic Index | 4.2 | 100.0 | Artificial Analysis official API |
Coding
| Benchmark | Score | Maximum | Methodology |
|---|---|---|---|
| LiveBench Coding | 60.3 | 100.0 | Contamination-free benchmark with objective ground-truth answers |
| AA Coding Index | 43.4 | 100.0 | Artificial Analysis official API |
| SciCode | 43.0 | 100.0 | — |
Data Analysis
| Benchmark | Score | Maximum | Methodology |
|---|---|---|---|
| LiveBench Data Analysis | 58.8 | 100.0 | Contamination-free benchmark with objective ground-truth answers |
Language
| Benchmark | Score | Maximum | Methodology |
|---|---|---|---|
| LiveBench Language | 71.3 | 100.0 | Contamination-free benchmark with objective ground-truth answers |
Long Context
| Benchmark | Score | Maximum | Methodology |
|---|---|---|---|
| AA-LCR | 62.0 | 100.0 | — |
Math
| Benchmark | Score | Maximum | Methodology |
|---|---|---|---|
| LiveBench Math | 73.9 | 100.0 | Contamination-free benchmark with objective ground-truth answers |
overall
| Benchmark | Score | Maximum | Methodology |
|---|---|---|---|
| LMArena Elo | 1452.0 | 2000.0 | Crowdsourced blind pairwise comparisons |
| LiveBench Global | 61.6 | 100.0 | Contamination-free benchmark with objective ground-truth answers |
| AA Intelligence Index | 14.7 | 100.0 | Artificial Analysis official API |
Reasoning
| Benchmark | Score | Maximum | Methodology |
|---|---|---|---|
| GPQA Diamond | 86.0 | 100.0 | Artificial Analysis official API |
| IFBench | 76.0 | 100.0 | — |
| LiveBench Reasoning | 59.4 | 100.0 | Contamination-free benchmark with objective ground-truth answers |
| HLE | 23.0 | 100.0 | — |
Tool Use
| Benchmark | Score | Maximum | Methodology |
|---|---|---|---|
| Tau²-Bench | 65.0 | 100.0 | — |
Information
- Release date
- April 02, 2026
- Tool Calling
- ✅ Supported
- Vision
- ✅ Supported
- Audio
- ❌ Not supported
Full Analysis: Gemma 4 31B
What is Gemma 4 31B?
Gemma 4 31B is an AI model developed by Google, classified as a large language model (LLM). It is a multimodal model, capable of processing text, images, and potentially other media types. As a proprietary model, it is available via Google's cloud API. With a context window of 262K tokens, it is suitable for processing long documents such as contracts, books, and complete codebases.
Pricing & Costs in 2026
Gemma 4 31B does not have public per-token pricing available at this time. Some models offer access via enterprise plans or research programs. Check Google's official website for up-to-date availability and pricing.
Benchmarks & Performance
Gemma 4 31B was evaluated on 17 different benchmarks, covering categories like Agentic, Coding, Data Analysis, Language, Long Context, Math, overall, Reasoning, Tool Use. Results show exceptional performance across available evaluations.
It's important to note that benchmarks measure specific aspects and don't capture the full user experience. Factors like instruction adherence, behavior in long conversations, and real-world task quality vary significantly between models and aren't always reflected in standard scores.
Recommended Use Cases
Gemma 4 31B is suitable for a wide range of AI applications: long document analysis (contracts, legal proceedings, codebases), automation with tool calling (API integration, databases, external systems), image and visual document analysis (OCR, diagrams, screenshots), multimodal processing combining text and images, complex reasoning, math problem solving, and logical analysis, text generation, summarization, translation, and general assistance.
Comparison with Alternatives
In the 2026 AI model ecosystem, Gemma 4 31B competes directly with similarly capable models. Key competitors include GPT (OpenAI), Claude (Anthropic), and open source models like Llama (Meta) and Qwen (Alibaba). The choice between models depends on the specific use case, budget, latency requirements, and need for features like multimodality and tool calling.
For a detailed side-by-side comparison, use our comparison tool or check the overall model ranking.
Frequently Asked Questions
What is 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 token context window, configurable thinking/reasoning mode, native function...
How much does Gemma 4 31B cost?
Gemma 4 31B does not have public per-token pricing available at this time. Check Google's official website for up-to-date information.
How does Gemma 4 31B compare with other models?
In available benchmarks, Gemma 4 31B scored: Terminal-Bench Hard: 36/100, AA Agentic Index: 4.2/100, LiveBench Coding: 60.33/100. See the full table above for a detailed comparison.
Is Gemma 4 31B open source?
No, Gemma 4 31B is a proprietary model from Google. It is available via cloud API. For open source alternatives, check our open source model ranking.
What is Gemma 4 31B best for?
Gemma 4 31B excels at complex reasoning and problem solving. With its large context window, it handles long documents, codebases, and extended conversations. It supports tool calling for API integrations and automation.
Last updated: October 16, 2026 • View methodology →