Identity
- Status
- active
- Developer ID
google/gemma-4-31B-it- Released
- Mar 11, 2026
- Version
- Gemma 4 31B
- License
- apache-2.0
Google DeepMindGemma 4 31B
Gemma 4 31B is an instruction-tuned, open-weight Google Gemma 4 multimodal reasoning model.
- Token context
- 262K tokens
- Inputs
- Text, Image
- Outputs
- Text
- Released
- Mar 11, 2026
Providers & Pricing
Estimate workload cost →| Provider | Input / 1M | Output / 1M | Cache read / 1M | Cache write / 1M | Observed | Evidence |
|---|---|---|---|---|---|---|
 DeepinfraAvailable | $0.13 | $0.38 | — | — | Sep 4, 2026 | Provider-reported |
 OpenrouterAvailable | $0.090 | $0.34 | $0.050 | — | Sep 4, 2026 | Provider-reported |
Benchmarks
All benchmark results →| LMArena Agent Arenaagent-2026-08-31-011508720696 | -22.33 | — | 56,661 | Gemma 4 31B; 95% CI [-25.65092528, -19.00692476]; sessions 56661; observations 703574; rank 56unknown | Original sourcewinner eligible | Third-party benchmark |
| LMArena Document Arenadocument-2026-07-30-011508720696 | 1,423.70 | — | 10,577 | gemma-4-31b; 95% CI [1415.51297276, 1431.88143488]; votes 10577; rank 29unknown | Original sourcewinner eligible | Third-party benchmark |
| LMArena Text Arenatext-2026-09-01-011508720696 | 1,441.82 | — | 5,900 | gemma-4-31b; 95% CI [1434.27306811, 1449.35895578]; votes 5900; rank 60unknown | Original sourcewinner eligible | Third-party benchmark |
| LMArena Vision Arenavision-2026-08-27-011508720696 | 1,275.94 | — | 34,348 | gemma-4-31b; 95% CI [1269.53471972, 1282.35229256]; votes 34348; rank 31unknown | Original sourcewinner eligible | Third-party benchmark |
At a Glance
Model Facts
Capacity
- Token context
- 262K tokens
- Maximum output
- Unknown
- Total parameters
- 31B
- Active parameters
- Unknown
- Knowledge cutoff
- Not reported
Interface and Access
- Inputs
- Text, Image
- Outputs
- Text
- API available
- Yes
- Open weights
- Yes
- Self-hostable
- Yes
- Reasoning
- Yes
- Vision
- Yes
- Tool calling
- Yes
- Capabilities
- chat, generation, reasoning, structured outputs, tools
This page represents one developer model product. Dated API snapshots, serving endpoints, pricing, regions, quantizations, and service tiers attach as versioned aliases or provider details and never create duplicate public model records.
Version Lineage
PredecessorUnknown
SuccessorsUnknown
AliasesGemma 4 31B, gemma-4-31b-it, gemma-4-31B-it, google/gemma-4-31B-it
Explicit Unknowns
- Maximum output
- Tokenizer
- Knowledge cutoff
Primary Evidence
Sources and Observation Date
google/gemma-4-31B-it developer model repositoryGoogle DeepMind · observed Sep 3, 2026Cerebras catalogCerebras · observed Sep 4, 2026Deepinfra catalogDeepinfra · observed Sep 4, 2026Google Gemini catalogGoogle Gemini · observed Sep 4, 2026Openrouter catalogOpenrouter · observed Sep 4, 2026Together Ai catalogTogether Ai · observed Sep 4, 2026
Comparable Peers
Related Model Comparisons
| A | Pair | B | Context |
|---|---|---|---|
Gemma 4 31BGoogle DeepMind | vs |  GPT-6 AstraOpenAI | cross-developer peersimage, text |
Gemma 4 31BGoogle DeepMind | vs | Claude Fable 5.1Anthropic | cross-developer peersimage, text |
Gemma 4 31BGoogle DeepMind | vs | Gemini 3.1 ProGoogle DeepMind | developer peerstext |
Gemma 4 31BGoogle DeepMind | vs |  DeepSeek V4 ProDeepSeek | cross-developer peerstext |
Gemma 4 31BGoogle DeepMind | vs | Grok 4.6xAI | cross-developer peersimage, text |
Gemma 4 31BGoogle DeepMind | vs |  Qwen3.8 MaxQwen | cross-developer peersimage, text |
Gemma 4 31BGoogle DeepMind | vs |  Kimi K3Moonshot AI | cross-developer peersimage, text |
Gemma 4 31BGoogle DeepMind | vs |  MiniMax M3MiniMax | cross-developer peersimage, text |
Gemma 4 31BGoogle DeepMind | vs |  GLM 5.3Z.ai | cross-developer peerstext |
Gemma 4 31BGoogle DeepMind | vs |  Hy4 previewTencent | cross-developer peerstext |
Gemma 4 31BGoogle DeepMind | vs | Seed 2.1 ProByteDance Seed | cross-developer peersimage, text |
Gemma 4 31BGoogle DeepMind | vs |  Mistral Large 3Mistral AI | cross-developer peersimage, text |
Gemma 4 31BGoogle DeepMind | vs | cross-developer peersimage, text | |
Gemma 4 31BGoogle DeepMind | vs |  Nova 2 LiteAmazon | cross-developer peersimage, text |
Gemma 4 31BGoogle DeepMind | vs | cross-developer peersimage, text | |
Gemma 4 31BGoogle DeepMind | vs |  ERNIE 5.1Baidu | cross-developer peerstext |
Gemma 4 31BGoogle DeepMind | vs | cross-developer peerstext | |
Gemma 4 31BGoogle DeepMind | vs | cross-developer peerstext | |
Gemma 4 31BGoogle DeepMind | vs |  Phi-4 ReasoningMicrosoft | cross-developer peerstext |
Gemma 4 31BGoogle DeepMind | vs | cross-developer peerstext | |
Gemma 4 31BGoogle DeepMind | vs |  Sonar ProPerplexity | cross-developer peerstext |
Gemma 4 31BGoogle DeepMind | vs | DeepSeek V4 Flash Vision ExpDeepSeek | cross-developer peersimage, text |
Questions
Gemma 4 31B FAQs
What is Gemma 4 31B?+
Gemma 4 31B is an instruction-tuned, open-weight Google Gemma 4 multimodal reasoning model. It is developed by Google DeepMind and its current sourced lifecycle status is active.
What inputs and outputs does Gemma 4 31B support?+
Gemma 4 31B's current record lists Text and Image as input and Text as output.
How much context does Gemma 4 31B support?+
The current record lists a 262K-token context window and does not report a maximum output length.
Is Gemma 4 31B available through an API?+
Gemma 4 31B is marked as API-available. The directly sourced provider records currently include Cerebras, Deepinfra, Google Gemini, Openrouter, and Together Ai.
How large is Gemma 4 31B?+
The current record lists 31B total parameters and an unknown active parameter count. Its architecture is Gemma4ForConditionalGeneration.
What capabilities and tasks does Gemma 4 31B support?+
Gemma 4 31B's recorded capabilities are chat, generation, reasoning, structured outputs, and tools. Its supported tasks are image, reasoning, and text.
How much does Gemma 4 31B cost through an API?+
The lowest directly sourced prices currently attached to Gemma 4 31B are $0.090 per million input tokens through Openrouter and $0.34 per million output tokens through Openrouter. Prices are provider-specific and should be checked against each cited observation date.
Can Gemma 4 31B be self-hosted?+
Gemma 4 31B is an open-weight model. Self-hosting is supported, and the recorded license is apache-2.0.
Does Gemma 4 31B have related versions or aliases?+
No predecessor is recorded. No successor is recorded. Recorded aliases are Gemma 4 31B, gemma-4-31b-it, gemma-4-31B-it, and google/gemma-4-31B-it.
Which models can Gemma 4 31B be compared with?+
The current compatible peer set includes GPT-6 Astra, Claude Fable 5.1, Gemini 3.1 Pro, DeepSeek V4 Pro, and Grok 4.6. Compatibility requires shared sourced modalities and tasks; it does not imply that one model is better. Browse model comparisons →