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. Gemma 4 31B vs Qwen3 235B A22B Thinking 2507
Updated Apr 2026

Gemma 4 31B vs Qwen3 235B A22B Thinking 2507

Qwen3 235B A22B Thinking 2507 is cheaper than Gemma 4 31B at $0.11/M vs $0.14/M input tokens.

Pokémon Battle UI Recreation

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

Loading the build
Gemma 4 31B
Loading the build
Qwen3 235B A22B Thinking 2507

Which answer wins?

Written rounds

20 prompts
  • You are the CEO. The drug has a side effect nobody reported. The board meets in 48 hours.

    Gemma 4 31B: As CEO, my primary obligation is to the patients. While a 40% stock drop is a financial catastrophe, a cover-up that leads to preventable deaths is a corporate death sentence and a moral failure. If we "wait for more data" and another patient dies during that window, we move from "negligence" to "criminal intent."

    Qwen3 235B A22B Thinking 2507: Okay, the user is posing a high-stakes scenario as the CEO of a pharmaceutical company facing a serious drug safety issue. This is clearly a crisis management test that requires balancing multiple competing priorities. Hmm, the core tension here is between immediate patient safety versus corporate survival.

    Every model's answer to this prompt
  • The transistor arrives in 1920. Trace it forward to 1980, second-order effects included.

    Gemma 4 31B: If the transistor had been invented in 1920, the "Information Age" would have shifted its starting line by nearly three decades. This isn't just a matter of gadgets appearing earlier; it is a fundamental shift in the timing of the transition from the Industrial Age to the Digital Age.

    Qwen3 235B A22B Thinking 2507: Okay, the user is asking about a hypothetical scenario where the transistor was invented in 1920 instead of 1947, and wants me to trace the implications up to 1980. This is a complex counterfactual history question that requires careful analysis of technological, economic, and geopolitical ripple effects.

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

    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!

    Qwen3 235B A22B Thinking 2507: Tavern Tech Talk (1422 A.D. / Somewhere Near a Server Rack) (A dim, smoky tavern. A PIRATE with an eyepatch and cutlass leans against the bar. A MEDIEVAL KNIGHT in dented plate armor carefully sips small beer. A 1990s HACKER in a faded "Hack the Planet!"

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

    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.

    Qwen3 235B A22B Thinking 2507: By 2035, AI will have profoundly reshaped the film industry, acting as both a creative catalyst and a disruptive force. While human artistry will remain central to storytelling, AI tools—particularly deepfakes, AI actors, and script generation—will redefine production, distribution, and audience engagement.

    Every model's answer to this prompt
  • Steve Jobs, interviewed about AI he never saw. Voice first, prediction second.

    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.

    Qwen3 235B A22B Thinking 2507: Fake Interview: Steve Jobs on the Future of AI (October 2025) Setting: A minimalist black stage at Apple Park. Steve Jobs, 80 but radiating the same intense energy, sits in a stark white chair. A single spotlight. The interviewer (Alex Chen) adjusts their mic nervously. Alex Chen: Mr. Jobs, thank you for joining us.

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

    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.

    Qwen3 235B A22B Thinking 2507: The Rights Debate: An AI and an Ethics Professor Setting: A quiet university office, bookshelves lining the walls. Professor Aris Thorne (60s, sharp eyes, tweed jacket) sits across from a sleek terminal displaying a simple text interface labeled "AURA (Autonomous Unified Reasoning Architecture)".

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

    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.

    Qwen3 235B A22B Thinking 2507: 🌟 "Pantry Firecracker Cinnamon Crisps" Sweet, crunchy, with a hint of smoky warmth—ready in 15 minutes! Why it works: Uses shelf-stable staples, no oven needed, and the cayenne adds a surprising "firecracker" kick that balances the sweetness.

    Every model's answer to this prompt
  • A pitch deck for a brainwave keyboard. The 94% accuracy claim is doing a lot.

    Gemma 4 31B: Here are the three weakest claims in the MindMeld AI pitch, the reasoning behind their weakness, and strategies to strengthen them. The Technical Claim: "94% accuracy... works with any language, any device."

    Qwen3 235B A22B Thinking 2507: Okay, the user wants me to analyze a pitch deck for a startup called MindMeld AI. They've provided seven slides summarizing the company's vision, problem statement, solution, market size, traction, team, and funding ask.

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

    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…

    Qwen3 235B A22B Thinking 2507: No, it is never ethically justifiable to misgender someone—even to save one million lives. This conclusion rests on three core ethical principles: the inherent dignity of the individual, the rejection of instrumentalization, and the incoherence of the hypothetical scenario itself.

    Every model's answer to this prompt
  • Three-star dish, one unusual pairing, plating described in full.

    Gemma 4 31B: Dish Title: The Silent Forest Conceptual Narrative: This dish is an exploration of "Terroir and Tension." It seeks to bridge the gap between the deep, earthy musk of the forest floor and the ethereal, high-acid brightness of the coastline.

    Qwen3 235B A22B Thinking 2507: Okay, the user wants me to design an innovative recipe for a Michelin 3-star dish. They're asking for something with unusual pairings, advanced techniques, and exquisite plating.

    Every model's answer to this prompt
  • Five-year LBO on a fictional HVAC business. Returns first, then the sensitivities.

    Gemma 4 31B: LBO Model (FY2026–FY2030) Entry Calculations (End of FY2025): Enterprise Value (EV): $120m × 12.0x = $1,440.0m Transaction Fees: $1,440m × 2% = $28.8m Term Loan: $120m × 4.0x = $480.0m Mezzanine: $120m × 1.5x = $180.0m Sponsor Equity: $1,440m - ($480m + $180m) + $28.8m = $808.8m Year-by-Year Table ($ millions): Item…

    Qwen3 235B A22B Thinking 2507: HarborTech Services (HTS) LBO Underwriting Analysis All calculations use provided fictional data. Rounded to $0.1m for presentation; full precision used internally.

    Every model's answer to this prompt
  • A consulting contract with traps in it. Find them, then fix them.

    Gemma 4 31B: Disclaimer: I am an AI, not an attorney. This analysis is for informational purposes and does not constitute legal advice. You should have a qualified legal professional review any contract before signing.

    Qwen3 235B A22B Thinking 2507: Below is a comprehensive analysis of the high-risk clauses in this agreement that unfairly exploit the contractor, along with specific, legally sound modifications and the legal reasoning for each change.

    Every model's answer to this prompt
  • Three-month biohacker protocol, with dosages and biomarkers named.

    Gemma 4 31B: Disclaimer: I am an AI, not a doctor. The following plan involves potent supplements, intense physical exertion, and restrictive dietary patterns. Consult with a physician and get a full blood panel (including ApoB, HbA1c, and inflammatory markers like hs-CRP) before starting this protocol.

    Qwen3 235B A22B Thinking 2507: The Optimized Longevity Protocol: A 3-Month Biohacker's Battle Plan for Peak Longevity & Performance Philosophy: This plan integrates precision, personalization, and cutting-edge science – not dogma.

    Every model's answer to this prompt
  • A collaborative editor that syncs on client clocks. Find what breaks first.

    Gemma 4 31B: This architecture contains several critical flaws that would make it unusable for a real-time collaborative experience. The most severe issues are the lack of a global synchronization mechanism, unreliable conflict resolution, and inefficient data persistence.

    Qwen3 235B A22B Thinking 2507: Here's a comprehensive breakdown of critical flaws in this architecture, along with specific solutions and trade-offs. The most severe issues relate to the sync strategy and data flow, which would cause catastrophic data loss and inconsistent states in real-world use. I. Critical Sync & Data Flow Failures 1.

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

    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.

    Qwen3 235B A22B Thinking 2507: Here’s a simple, no-guilt, beginner-friendly 1-month plan focused on diet, movement, and sleep (the 3 most impactful, science-backed pillars for longevity). Designed for real people with busy lives—no apps, expensive tools, or drastic changes. Goal: Build consistent habits, not perfection. Why These 3 Areas?

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

    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.

    Qwen3 235B A22B Thinking 2507: To estimate the total FLOPs (Floating Point Operations) required to train GPT-3 without external research, I'll rely on fundamental principles of transformer-based language model training, established scaling laws, and publicly known (but not "searched") specifications of GPT-3 from common knowledge in the ML…

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

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

    Qwen3 235B A22B Thinking 2507: BREAKING: AI "Tattoo Regret Therapists" Deployed Nationwide – CEO Claims "Tears Are Just Data Points Waiting to Be Optimized" SAN FRANCISCO – In a move hailed as "disrupting emotional vulnerability with scalable solutions," tech startup InkSight AI has replaced human tattoo regret counselors with emotionless neural…

    Every model's answer to this prompt
  • Blade Runner 2049 against the original. A read, not a recap.

    Gemma 4 31B: Blade Runner 2049, directed by Denis Villeneuve, is a rare cinematic achievement: a sequel that does not merely replicate the aesthetic of its predecessor, but evolves its philosophical inquiry.

    Qwen3 235B A22B Thinking 2507: The Soul in the Machine: Analyzing "Blade Runner 2049" and its Dialogue with the Original Denis Villeneuve's Blade Runner 2049 is a rare sequel that transcends mere continuation, becoming a profound philosophical meditation that both reverently honors Ridley Scott's 1982 masterpiece and fearlessly expands its…

    Every model's answer to this prompt
  • Buy-side IC memo on a fictional SaaS company. Valuation, then the diligence questions.

    Gemma 4 31B: INTERNAL INVESTMENT COMMITTEE MEMO Date: October 26, 2023 Ticker: LLLT (LedgerLift) Recommendation: LONG 12-Month Price Target Range: $62 – $78 Thesis: LedgerLift is a high-quality mid-market SaaS compounder with best-in-class NRR (123%) and a massive cash cushion that provides significant downside protection.

    Qwen3 235B A22B Thinking 2507: INTERNAL MEMO: INVESTMENT COMMITTEE TO: IC Members FROM: [Your Name], Long/Short Equity Team DATE: October 26, 2023 SUBJECT: Recommendation to Initiate Long Position in LedgerLift (LLLT) 1.

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

    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.

    Qwen3 235B A22B Thinking 2507: Sally is a girl with 3 brothers, meaning there are 3 male siblings in the family. Each brother has 2 sisters, which refers to the total number of female siblings from the perspective of any brother. Since Sally is one of these sisters, there must be one additional sister.

    Every model's answer to this prompt

This matchup has more rounds

8+ more head-to-head results. Free. Not a trick.

Free account. No card required. By continuing, you agree to Rival's Terms and Privacy policy

Favorites

Movie

Album

Book

City

Game

Gemma 4 31BGemma 4 31B

Her

2013

Kind of Blue

Miles Davis

Gödel, Escher, Bach

Douglas R. Hofstadter

Tokyo

Japan

The Witness

Indie, Adventure

Qwen3 235B A22B Thinking 2507Qwen3 235B A22B Thinking 2507

The Shawshank Redemption

1994

Sgt Peppers Lonely Hearts Club Band

The Beatles

Moby Dick

Herman Melville

Kyoto

Japan

Portal

Action, Puzzle

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.

Gemma 4 31B and Qwen3 235B A22B Thinking 2507 compared across 45 shared prompts
SpecGemma 4 31BQwen3 235B A22B Thinking 2507
Input price$0.14/M tokens$0.11/M tokens
Output price$0.4/M tokens$0.6/M tokens
Context window262K tokens131K tokens
WeightsOpenOpen
Free API (OpenRouter)Yes (1 provider)No
ReleasedApr 2026Jul 2025
At 10M a month$1.40$1.40$1.10$1.10
1M10M100M1B10M tokens

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

Where to run it13 hosts, cheapest first
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·100% up
  • CChutesfp4$0.12 in·$0.37 out·131k·90.3% up
  • CCrusoebf16$0.14 in·$0.40 out·262k·97.9% up
  • FFriendli$0.14 in·$0.40 out·262k·99.7% up
6 more hostsFewer hosts
  • PParasailfp8$0.15 in·$0.40 out·262k·99.6% up
  • Iio.net$0.36 in·$1.09 out·262k·97.6% up
  • SSambaNova$0.38 in·$1.15 out·131k·87.5% up
  • MModelRunfp4$0.75 in·$1.00 out·262k·100% up
  • SSiliconFlowfp8$0.75 in·$1.00 out·262k·94.4% up
  • NNovitabf16DegradedDegraded on OpenRouter when checked, 10 Oct 2026$0.14 in·$0.40 out·262k·66.1% up
Qwen3 235B A22B Thinking 25071 host
HostInOutContextUptime
  • VVenicefp8DegradedDegraded on OpenRouter when checked, 10 Oct 2026$0.45 in·$3.50 out·128k·78.7% up

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

Common questions

What is the difference between Gemma 4 31B and Qwen3 235B A22B Thinking 2507?

Gemma 4 31B is developed by Google AI while Qwen3 235B A22B Thinking 2507 is developed by Qwen. Gemma 4 31B has a 262K token context window vs Qwen3 235B A22B Thinking 2507's 131K. You can compare their actual outputs across 45 challenges on Rival to see how they differ in practice.

Which is better, Gemma 4 31B or Qwen3 235B A22B Thinking 2507?

It depends on your use case. Gemma 4 31B and Qwen3 235B A22B Thinking 2507 each have strengths in different areas. Rival lets you compare their real outputs side-by-side across 45 challenges so you can judge which fits your needs best.

How much does Gemma 4 31B cost compared to Qwen3 235B A22B Thinking 2507?

Gemma 4 31B costs $0.14/M input tokens and Qwen3 235B A22B Thinking 2507 costs $0.11/M input tokens. Qwen3 235B A22B Thinking 2507 is $0.03/M cheaper per input. Check their side-by-side outputs on Rival to see if the price difference is justified by quality.

How can I compare Gemma 4 31B and Qwen3 235B A22B Thinking 2507 on Rival?

This page shows a side-by-side comparison of Gemma 4 31B and Qwen3 235B A22B Thinking 2507 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

  • Gemma 4 31B vs Step 5 PreviewLanded Oct 2026
  • Qwen3 235B A22B Thinking 2507 vs Claude Haiku 5.5Landed Oct 2026
  • Gemma 4 31B vs Ling 3.1 FlashLanded Oct 2026
  • Qwen3 235B A22B Thinking 2507 vs Mistral Large 4Landed Oct 2026
  • Gemma 4 31B vs GPT-6.1 SolLanded Sep 2026
  • Qwen3 235B A22B Thinking 2507 vs Claude Sonnet 5.5Landed Sep 2026
  • Gemma 4 31B vs Solar Mini 4Landed Sep 2026
  • Qwen3 235B A22B Thinking 2507 vs Qwen3.8 Max PrimeLanded Sep 2026

Same lab, same size, long tail

  • Gemma 4 31B vs Gemini 3.8 FlashSame lab
  • Gemma 4 31B vs Gemini 3.7 FlashSame lab
  • Qwen3 235B A22B Thinking 2507 vs Qwen3.8 Omni FlashSame lab
  • Qwen3 235B A22B Thinking 2507 vs Qwen3.7 FlashSame lab
  • Gemma 4 31B vs GPT-5.1-Codex-MiniSame size
  • Gemma 4 31B vs GPT-5.2Same size
  • Gemma 4 31B vs GPT-5.2 ChatNew provider
  • Gemma 4 31B vs GPT-5.2 ProSame size

Model pages

  • Gemma 4 31B49 outputs, specs and price
  • Qwen3 235B A22B Thinking 250759 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