Gemma 4 31B vs GPT-5.4 mini
Gemma 4 31B and GPT-5.4 mini are closely matched at 33.1 and 30.1 on the LLM Stats Score. Gemma 4 31B is 11.1x cheaper per token.
Google · OpenAI · Updated for 2026
Which is better?
Gemma 4 31B and GPT-5.4 mini are closely matched on the overall LLM Stats Score at 33.1 and 30.1.
The models split the 4 individual benchmarks reported for both models evenly.
On price, Gemma 4 31B is roughly 11.1x cheaper per token on a blended 3:1 input/output basis, which adds up quickly at production volume.
GPT-5.4 mini also accepts a larger context window (400,000 input tokens), making it the stronger choice for long documents and large codebases.
Based on current LLM Stats indexes, shared benchmarks, pricing, and model metadata for 2026.
Choose Gemma 4 31B
- cost matters — it's about 11.1x cheaper per token
- you want the most recent training data — it shipped Apr 2026
- you need open weights you can self-host or fine-tune
Choose GPT-5.4 mini
- you process long inputs — it offers a 400,000 token context window
At a glance
The differences that matter most.
Capability indexes
Additional strengths measured across groups of related public benchmarks
Individual benchmarks
12 reported for Gemma 4 31B · 15 for GPT-5.4 mini
Gemma 4 31B outperforms in 2 benchmarks (MMMU-Pro, MRCR v2 (8-needle)), while GPT-5.4 mini is better at 2 benchmarks (GPQA, Humanity's Last Exam).
Both models are evenly matched across the benchmarks.
Human preference
Blind head-to-head votes and playground preference scores
Pricing Analysis
Price comparison per million tokens
For input processing, Gemma 4 31B ($0.09/1M tokens) is 8.3x cheaper than GPT-5.4 mini ($0.75/1M tokens).
For output processing, Gemma 4 31B ($0.34/1M tokens) is 13.2x cheaper than GPT-5.4 mini ($4.50/1M tokens).
In conclusion, GPT-5.4 mini is more expensive than Gemma 4 31B.*
* Using a 3:1 ratio of input to output tokens
Context Window
Maximum input and output token capacity
GPT-5.4 mini accepts 400,000 input tokens compared to Gemma 4 31B's 262,144 tokens. Gemma 4 31B can generate longer responses up to 262,144 tokens, while GPT-5.4 mini is limited to 128,000 tokens.
Input capabilities
Documented input modalities across available providers
Both Gemma 4 31B and GPT-5.4 mini support multimodal inputs.
They are both capable of processing various types of data, offering versatility in application.
Gemma 4 31B
GPT-5.4 mini
License
Usage and distribution terms
Gemma 4 31B is licensed under Apache 2.0, while GPT-5.4 mini uses a proprietary license.
License differences may affect how you can use these models in commercial or open-source projects.
Apache 2.0
Open weights
Proprietary
Closed source
Release Timeline
When each model was launched
Gemma 4 31B was released on 2026-04-02, while GPT-5.4 mini was released on 2026-03-17.
Gemma 4 31B is 1 month newer than GPT-5.4 mini.
Apr 2, 2026
5 months ago
2w newerMar 17, 2026
6 months ago
Knowledge Cutoff
When training data ends
Gemma 4 31B has a knowledge cutoff of 2025-01-01, while GPT-5.4 mini has a cutoff of 2025-08-31.
GPT-5.4 mini has more recent training data (up to 2025-08-31), making it potentially better informed about events through that date compared to Gemma 4 31B (2025-01-01).
Jan 2025
Aug 2025
7 mo newerProvider Availability
Gemma 4 31B is available from DeepInfra, FriendliAI, Novita, Together. GPT-5.4 mini is available from OpenAI.
Gemma 4 31B
GPT-5.4 mini
Outputs Comparison
Judge for yourself.
Run your own prompts against Gemma 4 31B and GPT-5.4 mini side-by-side, then vote on the output you prefer.
FAQ
Common questions about Gemma 4 31B vs GPT-5.4 mini.