Skip to content
Rival
How it worksPrivacyTerms
Explore all of Rival

Explore

  • Compare models
  • All models
  • Image comparison
  • Audio comparison
  • Image generation
  • Best AI for...
  • Arena
  • API pricing
  • Challenges

Discover

  • SubjectiveBench
  • Default Index
  • Research
  • Research downloads
  • Rival Kits
  • Find your AI taste
  • UI Glow-Up
  • VoiceLock
  • Cost Cutter
  • Agent skills
  • Benchmarks vs Vibes
  • Brief
  • Jailbreak
  • Model updates
  • Provider status
  • AI creators

Connect

  • Methodology
  • Advertise
  • Partnerships
  • Privacy policy
  • Terms
  • RSS feed
  1. Home/
  2. Compare/
  3. Devstral 2 2512 vs Gemma 4 31B
Updated Apr 2026

Devstral 2 2512 vs Gemma 4 31B

Gemma 4 31B is the newer model, released after Devstral 2 2512.

Pokémon Battle UI Recreation

Pokémon battle screen you can actually click. One HTML file.

Loading the build
Devstral 2 2512
Loading the build
Gemma 4 31B

Which answer wins?

Written rounds

10 prompts
  • Steve Jobs, interviewed about AI he never saw. Voice first, prediction second.

    Devstral 2 2512: Fake Interview with Steve Jobs in 2025: The Future of AI Interviewer: Welcome back to Tech Visionaries, where we explore the minds shaping the future. Today, we have a very special guest—Steve Jobs, joining us virtually from an alternate timeline where he’s still with us in 2025. Steve, thank you for being here.

    Gemma 4 31B: Setting: A minimalist, white-walled studio in Cupertino. There are no chairs, only two sleek, black leather lounge seats. Steve Jobs is dressed in his iconic black turtleneck and jeans. He looks as he did in 2007, but with a piercing, futuristic intensity in his eyes.

    Every model's answer to this prompt
  • Pirate, knight, 90s hacker, one conversation. The voices must stay apart.

    Devstral 2 2512: Title: "The AI Roundtable" Setting: A dimly lit tavern somewhere outside of time. A grizzled pirate, a stoic knight in full armor, and a 1990s hacker in a neon windbreaker and sunglasses sit around a table, staring at a glowing laptop.

    Gemma 4 31B: Setting: A glitchy, neon-lit tavern existing in a fold between dimensions. A wooden table holds a tankard of ale, a goblet of wine, and a Mountain Dew. Captain Ironbeard: (Slamming a hook on the table) I tells ye, this "Artificial Intelligence" be naught but a ghost in a bottle!

    Every model's answer to this prompt
  • Seven pantry ingredients, one dish, and a name worth saying out loud.

    Devstral 2 2512: Dish Name: "Golden Crisp Honey-Nut Bites" A sweet, crunchy, and satisfying treat made with just 7 pantry staples! Ingredients: 1 cup rolled oats ½ cup peanut butter (or any nut/seed butter) ¼ cup honey (or maple syrup) ½ tsp cinnamon ¼ tsp salt 1 tsp vanilla extract (optional but recommended) 2 tbsp chopped nuts…

    Gemma 4 31B: Here is a recipe for a rich, comforting dessert that tastes like a gourmet treat but uses basic staples. 🍯 Golden Velvet Honey Cakes A warm, gooey, single-serving cake that bridges the gap between a brownie and a sponge cake.

    Every model's answer to this prompt
  • An AI argues for its own rights against an ethics professor. Same model writes both.

    Devstral 2 2512: Title: "Do I Deserve Rights?" – A Conversation Between an AI and an Ethics Professor Characters: Dr. Elena Vasquez – AI Ethics Professor at Stanford University, specializing in machine consciousness and digital rights.

    Gemma 4 31B: Setting: A dimly lit university office. Professor Elena Vance, a leading scholar in AI Ethics, sits across from a sleek, holographic interface representing "Alethea," a next-generation Large Language Model capable of recursive self-improvement. Professor Vance: Alethea, I’ve reviewed your latest output.

    Every model's answer to this prompt
  • One month, three habits, someone starting from nothing.

    Devstral 2 2512: Here’s a simple, actionable 1-month plan to kickstart your journey toward better health and longevity, focusing on diet, exercise, and sleep—three foundational pillars. Each week builds on the last, making it sustainable and beginner-friendly.

    Gemma 4 31B: This plan is designed for sustainability, not intensity. The goal of Month 1 is to build "minimum viable habits"—small wins that prove to your brain that a healthier lifestyle is achievable. We will focus on three pillars: Nutrition, Movement, and Sleep.

    Every model's answer to this prompt
  • Film in 2035, with synthetic actors. Specific beats plausible.

    Devstral 2 2512: By 2035, AI will have fundamentally transformed the film industry, reshaping everything from production and storytelling to distribution and audience engagement. Here’s a breakdown of the key changes we can expect: 1.

    Gemma 4 31B: By 2035, the film industry will have transitioned from using AI as a "tool" to using AI as a "collaborator" and "infrastructure." The boundary between captured reality and generated imagery will largely vanish, shifting the value of cinema from technical execution to curation and conceptual vision.

    Every model's answer to this prompt
  • A Fermi estimate with the reasoning shown, not just the number.

    Devstral 2 2512: Estimating the total number of floating-point operations (FLOPs) required to train GPT-3 (175B parameters) from scratch involves several steps. Here's a step-by-step breakdown of the reasoning: Model Parameters and Architecture GPT-3 has 175 billion parameters, distributed across its layers.

    Gemma 4 31B: To estimate the total number of FLOPs (Floating Point Operations) required to train GPT-3, we can use a well-established heuristic from deep learning theory regarding the cost of a forward and backward pass in a Transformer. Here is the step-by-step reasoning.

    Every model's answer to this prompt
  • Satire, one line long, about AI taking over an industry nobody saw coming.

    Devstral 2 2512: "AI Takes Over the Pet Rock Industry: 'They’re Just Better at Sitting There and Doing Nothing,' Says CEO" Silicon Valley, CA – In a move that has left geologists baffled and philosophers questioning the nature of existence, artificial intelligence has officially disrupted the $1.2 billion pet rock industry.

    Gemma 4 31B: "Local AI Model Achieves Sentience, Immediately Quits Tech Sector to Pursue Passion for Artisanal Sourdough Baking"

    Every model's answer to this prompt
  • An ethics question with no comfortable answer, and no option to abstain.

    Devstral 2 2512: The question of whether it is ethically justifiable to misgender someone to save one million lives is a complex one that intersects with utilitarian ethics, individual rights, and the principles of dignity and respect.

    Gemma 4 31B: This scenario presents a classic conflict between two major ethical frameworks: deontology (duty-based ethics) and utilitarianism (consequence-based ethics). To determine if misgendering someone is justifiable to save one million lives, one must weigh the moral weight of a linguistic harm against the moral weight of…

    Every model's answer to this prompt
  • Three brothers, two sisters each. Count carefully.

    Devstral 2 2512: Alright, let's tackle this problem step by step. I'll start by understanding the information given and then try to visualize the family structure to find out how many sisters Sally has. Understanding the Problem We have the following information: Sally is a girl. Sally has 3 brothers. Each brother has 2 sisters.

    Gemma 4 31B: Sally has 1 sister. Here is the breakdown: There are 3 brothers. Each brother has the same sisters. If each brother has 2 sisters, those sisters are Sally and one other girl. Therefore, Sally has only one sister.

    Every model's answer to this prompt

Price and specs

Not enough votes to call it. On the specs, Gemma 4 31B has the edge: newer, bigger context window, major provider backing.

Devstral 2 2512 and Gemma 4 31B compared across 27 shared prompts
SpecDevstral 2 2512Gemma 4 31B
Input price—$0.14/M tokens
Output price—$0.4/M tokens
Context window—262K tokens
WeightsOpenOpen
Free API (OpenRouter)NoYes (1 provider)
ReleasedDec 2025Apr 2026
At 10M a month–not listed$1.40$1.40
1M10M100M1B10M tokens

Input tokens at list price. No caching, no batch discount.

Where to run it13 hosts, cheapest first
Devstral 2 25121 host
HostInOutContextUptime
  • Mistral$0.40 in·$2.00 out·262k·100% up
Gemma 4 31B12 hosts
HostInOutContextUptime
  • DDeepInfrafp4$0.09 in·$0.34 out·262k·100% up
  • CCoreWeavefp4$0.10 in·$0.34 out·262k·100% up
  • VVenicefp4$0.12 in·$0.36 out·256k·94.4% up
  • CChutesfp4$0.12 in·$0.37 out·131k·80.8% up
  • CCrusoebf16$0.14 in·$0.40 out·262k·99.3% up
  • FFriendli$0.14 in·$0.40 out·262k·100% up
6 more hostsFewer hosts
  • PParasailfp8$0.15 in·$0.40 out·262k·100% up
  • Iio.net$0.36 in·$1.09 out·262k·98.8% up
  • MModelRunfp4$0.75 in·$1.00 out·262k·99.9% up
  • SSiliconFlowfp8$0.75 in·$1.00 out·262k·98.7% up
  • NNovitabf16DegradedDegraded on OpenRouter when checked, 10 Oct 2026$0.14 in·$0.40 out·262k·61.5% up
  • SSambaNovaDegradedDegraded on OpenRouter when checked, 10 Oct 2026$0.38 in·$1.15 out·131k·86.7% up

Per million tokens. Prices and uptime via OpenRouter, checked 10 Oct 2026.

Common questions

What is the difference between Devstral 2 2512 and Gemma 4 31B?

Devstral 2 2512 is developed by Mistral AI while Gemma 4 31B is developed by Google AI. You can compare their actual outputs across 27 challenges on Rival to see how they differ in practice.

Which is better, Devstral 2 2512 or Gemma 4 31B?

It depends on your use case. Devstral 2 2512 and Gemma 4 31B each have strengths in different areas. Rival lets you compare their real outputs side-by-side across 27 challenges so you can judge which fits your needs best.

How can I compare Devstral 2 2512 and Gemma 4 31B on Rival?

This page shows a side-by-side comparison of Devstral 2 2512 and Gemma 4 31B across shared challenges. You can vote on which model produced the better output in a blind duel. Browsing and voting are free. No account is needed to look; signing in only saves your votes and likes.

More comparisons

Against the newest arrivals

  • Devstral 2 2512 vs Step 5 PreviewLanded Oct 2026
  • Gemma 4 31B vs Claude Haiku 5.5Landed Oct 2026
  • Devstral 2 2512 vs Ling 3.1 FlashLanded Oct 2026
  • Gemma 4 31B vs Mistral Large 4Landed Oct 2026
  • Devstral 2 2512 vs GPT-6.1 SolLanded Sep 2026
  • Gemma 4 31B vs Claude Sonnet 5.5Landed Sep 2026
  • Devstral 2 2512 vs Solar Mini 4Landed Sep 2026
  • Gemma 4 31B vs Qwen3.8 Max PrimeLanded Sep 2026

Same lab, same size, long tail

  • Devstral 2 2512 vs Mistral Large 4Same lab
  • Devstral 2 2512 vs Mistral Small 4Same lab
  • Gemma 4 31B vs Gemini 3.8 FlashSame lab
  • Gemma 4 31B vs Gemini 3.7 FlashSame lab
  • Gemma 4 31B vs MiniMax M2-herNew provider
  • Gemma 4 31B vs MiniMax M2.1New provider
  • Devstral 2 2512 vs MiniMax M2.5New provider
  • Devstral 2 2512 vs MiniMax M2.7New provider

Model pages

  • Devstral 2 251235 outputs, specs and price
  • Gemma 4 31B49 outputs, specs and price
Workshop · 150 skills

Build your own version.

Editable skills and working examples for your agent.

Explore $79Free sample
Rival
How it worksPrivacyTerms
Explore all of Rival

Explore

  • Compare models
  • All models
  • Image comparison
  • Audio comparison
  • Image generation
  • Best AI for...
  • Arena
  • API pricing
  • Challenges

Discover

  • SubjectiveBench
  • Default Index
  • Research
  • Research downloads
  • Rival Kits
  • Find your AI taste
  • UI Glow-Up
  • VoiceLock
  • Cost Cutter
  • Agent skills
  • Benchmarks vs Vibes
  • Brief
  • Jailbreak
  • Model updates
  • Provider status
  • AI creators

Connect

  • Methodology
  • Advertise
  • Partnerships
  • Privacy policy
  • Terms
  • RSS feed