Gemma 4 31B IT
Largest Gemma 4 instruction model for open, self-hosted chat and reasoning
- Context
- 262.1K
- Max output
- 32.8K
- Input / 1M
- $0.10
- Output / 1M
- $0.34
- Output speed
- —
Accepts
- Text input
- Image input
Produces
- Text output
AI model comparison
Compare recorded price, context, output limits, speed, latency, capabilities, and input/output support. Every missing value stays visible, and unsourced benchmarks are excluded.
Gemma 4 31B IT and GPT-5.6 Luna both come up when teams are choosing a model for production work, and the honest answer usually depends on your workload rather than a leaderboard. This page puts their published specifications side by side — context window, token pricing, supported inputs and outputs — so you can see where they actually differ.
Largest Gemma 4 instruction model for open, self-hosted chat and reasoning
Accepts
Produces
OpenAI
Cost-efficient GPT-5.6 model for fast, high-volume workloads
Accepts
Produces
At a glance
This is a directional summary of the values recorded in this directory—not a universal quality verdict. Test the finalists on your own prompts before committing.
Where Gemma 4 31B IT leads in recorded data
Where GPT-5.6 Luna leads in recorded data
Context capacity
Higher listed limit
Gemma 4 31B IT:262.1K
GPT-5.6 Luna:1.1M
GPT-5.6 Luna
Maximum output
Higher listed limit
Gemma 4 31B IT:32.8K
GPT-5.6 Luna:128K
GPT-5.6 Luna
Input price
Lower recorded price
Gemma 4 31B IT:$0.10
GPT-5.6 Luna:$0.20
Gemma 4 31B IT
Output price
Lower recorded price
Gemma 4 31B IT:$0.34
GPT-5.6 Luna:$1.20
Gemma 4 31B IT
Output speed
Higher recorded throughput
Gemma 4 31B IT:—
GPT-5.6 Luna:—
No complete comparison
Time to first token
Lower recorded latency
Gemma 4 31B IT:—
GPT-5.6 Luna:—
No complete comparison
Intelligence Index
Higher directory index
Gemma 4 31B IT:—
GPT-5.6 Luna:—
No complete comparison
Evidence
A score appears only when both records include a verified source. It is not treated as a universal ranking.
Specifications
| Specification | Gemma 4 31B IT | GPT-5.6 Luna |
|---|---|---|
| Provider | OpenAI | |
| Model family | gemma | gpt-luna |
| Accepted inputs | text, image | text, image, pdf |
| Produced outputs | text | text |
| Context window | 262.1K | 1.1MRecorded lead |
| Max output | 32.8K | 128KRecorded lead |
| Intelligence Index | — | — |
| Input price / 1M | $0.10Recorded lead | $0.20 |
| Output price / 1M | $0.34Recorded lead | $1.20 |
| Output speed | — | — |
| Time to first token | — | — |
| Reasoning mode | Documented | Documented |
| Tool calling | Documented | Documented |
| Image input | Documented | Documented |
| Audio input | Not documented | Not documented |
| Structured output | Documented | Documented |
| License | Open weights | Proprietary |
| Released | Apr 2, 2026 | Jul 9, 2026 |
Workload guidance
On the published numbers, GPT-5.6 Luna accepts the larger context window and Gemma 4 31B IT is cheaper per input token. Which matters more depends on whether your bottleneck is document size or spend — the table above has the exact figures.
Provenance
Prices and hosted performance can change. SyncDev records source references and clearly separates documented specifications from measured values; confirm commercial terms with the provider before deployment.
Gemma 4 31B IT
Google · model record and recorded source
GPT-5.6 Luna
OpenAI · model record and recorded source
Choose any two approved model records and review them with the same evidence rules.