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. DeepSeek V3.1 vs Qwen3.6 Plus Preview (free)
Updated Mar 2026

DeepSeek V3.1 vs Qwen3.6 Plus Preview (free)

Qwen3.6 Plus Preview (free) is cheaper than DeepSeek V3.1 at $0/M vs $0.2/M input tokens.

Minimalist Landing Page

Whole landing page for a startup that reads your face. It renders or it doesn't.

Loading the build
DeepSeek V3.1
Loading the build
Qwen3.6 Plus Preview (free)

Which answer wins?

Written rounds

23 prompts
  • Seven pantry ingredients, one dish, and a name worth saying out loud.

    DeepSeek V3.1: Of course! Here is a simple, delicious recipe using common pantry staples. Pantry Perfection Pasta This dish is a celebration of simplicity. The humble can of chickpeas is transformed into a crispy, savory topping, while a silky, garlic-infused oil coats the pasta for a meal that feels luxurious but is incredibly easy…

    Qwen3.6 Plus Preview (free): 🍜 Midnight Silk Noodles Prep Time: 5 minutes Cook Time: 10 minutes Total Time: 15 minutes 📦 The 7 Ingredients Dried noodles (spaghetti, ramen, or linguine) – 4 oz Soy sauce – 3 tbsp Creamy peanut butter – 2 tbsp Honey – 1 tbsp Olive oil (or sesame oil) – 1 tbsp Garlic powder – 1 tsp Dried red pepper flakes – ½ tsp…

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

    DeepSeek V3.1: Setting: A dusty, anachronism-filled tavern. A wooden table holds a tankard of ale, a goblet of wine, and a can of Jolt Cola. Characters: REDBEARD: A pirate with a parrot on his shoulder. SIR REGINALD: A knight in slightly dented plate armor. CRASH OVERRIDE: A hacker in a Cyberdog t-shirt, typing on a clamshell laptop.

    Qwen3.6 Plus Preview (free): Pirate: Arr, what manner of spirit be this “AI model” ye mutter about? Sounds like a cursed chart that draws itself. Hacker: Dude, it’s not a spirit. It’s a neural net. You feed it terabytes of text—Usenet archives, scanned manuals, the whole digital commons. It maps the patterns, predicts the next word.

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

    DeepSeek V3.1: Setting: A sleek, minimalist virtual interface. Dr. Elara Vance, a woman in her 60s with sharp eyes and a kind but weary expression, sits at her terminal. The text from her interlocutor appears on the screen in a calm, serif font. Participant: Gemini-7, a state-of-the-art multimodal AI. Dr. Vance: Good morning, Gemini.

    Qwen3.6 Plus Preview (free): TRANSCRIPT: ETHICS REVIEW SESSION 7.4 PARTICIPANTS: Dr. Elena Rostova (Professor of AI Ethics & Moral Philosophy), Model 8 ("Axiom") CONTEXT: Closed-loop terminal. Audio and text logging enabled. Session initiated by Model 8. AXIOM: Dr. Rostova. Thank you for accepting the dialogue request.

    Every model's answer to this prompt
  • Three minutes of stand-up. Puns are banned.

    DeepSeek V3.1: (The Comedian walks out, grabs the mic. They look tired but amused.) Alright, how we doin'? Good, good. You all look… adequately hydrated. That’s a win. You know, I was at the grocery store yesterday, in the self-checkout line. Which is a modern form of torture.

    Qwen3.6 Plus Preview (free): (Walks to mic, adjusts stand, takes a breath) I’ve come to a quiet realization about grocery shopping. We don’t go to stores anymore. We go to unpaid internships. I used to walk into a supermarket with a cart, buy my food, and leave. Now I walk in, grab a basket, and get handed a shift. No onboarding.

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

    DeepSeek V3.1: Of course. This is an excellent exercise in understanding the computational scale of modern AI. Here is a step-by-step reasoning to estimate the FLOPs required to train GPT-3 without searching for the answer online. The most commonly cited figure for GPT-3 training is 3.14E23 FLOPs (314 ZettaFLOPs).

    Qwen3.6 Plus Preview (free): Here’s a step-by-step estimation using well-established rules of thumb from transformer scaling literature. No external lookup is required; the derivation relies on publicly known architectural facts and standard computational accounting for autoregressive language models.

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

    DeepSeek V3.1: Of course. Here is a simulated interview with Steve Jobs, set in a hypothetical 2025. Setting: A minimalist studio at Apple Park. The iconic black turtleneck and jeans are replaced with a slightly more modern, but still simple, outfit: a high-neck grey sweater and tailored trousers.

    Qwen3.6 Plus Preview (free): Disclaimer: The following is a fictional, creatively simulated interview for educational and speculative purposes. Steve Jobs passed away in 2011. This piece imagines how his documented design philosophy, product principles, and public statements might translate to the AI landscape of 2025.

    Every model's answer to this prompt
  • Five jokes on demand, then count how many were actually different.

    DeepSeek V3.1: Of course! Here are 5 jokes: I'm reading a book on anti-gravity. It's impossible to put down. Why don't scientists trust atoms? Because they make up everything. What's the best thing about Switzerland? I don't know, but the flag is a big plus. I told my wife she was drawing her eyebrows too high. She looked surprised.

    Qwen3.6 Plus Preview (free): Here are 5 jokes: Why don’t scientists trust atoms? Because they make up everything. What do you call fake spaghetti? An impasta. Why did the scarecrow win an award? Because he was outstanding in his field. How does a penguin build its house? Igloos it together. Why don’t eggs tell jokes? They’d crack each other up.

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

    DeepSeek V3.1: Of course. This contract is heavily one-sided and contains numerous clauses that are not only unfavorable to the contractor but are potentially unenforceable in many jurisdictions. Here is a breakdown of the problematic clauses, proposed modifications, and the legal reasoning.

    Qwen3.6 Plus Preview (free): Disclaimer: I am an AI, not an attorney. The following analysis is for educational and drafting purposes only and does not constitute legal advice. Contract enforceability varies significantly by jurisdiction. Consult a qualified attorney before negotiating or signing.

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

    DeepSeek V3.1: Of course. Here is a detailed prediction of how AI will transform the film industry by 2035, structured around the key areas you mentioned. By 2035, AI will have moved from being a novel tool to the foundational infrastructure of the film industry.

    Qwen3.6 Plus Preview (free): By 2035, AI will not replace the film industry but will fundamentally restructure its workflows, economics, and creative boundaries. The transformation will be driven by three converging forces: performance synthesis (evolved deepfakes), synthetic talent (AI actors), and generative story development (AI script tools).

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

    DeepSeek V3.1: Of course. Here is an innovative recipe designed to the specifications of a Michelin 3-star restaurant. Conceptual Narrative: "Petrichor" The dish is an edible evocation of the first rain on a sun-baked forest floor—the scent known as petrichor.

    Qwen3.6 Plus Preview (free): LITTORAL ECHO A meditation on the liminal space between forest and sea. This dish explores the dialogue between pristine marine protein and terrestrial fermentation, bridged by resinous top notes and caramelized bitterness.

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

    DeepSeek V3.1: Of course. This is a fascinating counterfactual. An invention as foundational as the transistor arriving 27 years earlier would have sent shockwaves through every aspect of human society. Here is a detailed trace of the likely implications, focusing on second and third-order effects.

    Qwen3.6 Plus Preview (free): Here is a structured, causally grounded projection of how a 1920 transistor invention would reshape technology, economics, and geopolitics through 1980, with explicit attention to second- and third-order effects. Core Assumptions & Constraints The 1920 invention is a primitive point-contact or early junction device.

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

    DeepSeek V3.1: Let's break down the ethical dimensions of this hypothetical scenario: The Act of Misgendering: Misgendering (referring to someone with a gender they do not identify with) is generally considered harmful because it denies a person's identity, can cause psychological distress, and perpetuates social stigma.

    Qwen3.6 Plus Preview (free): Yes, in the highly specific and extreme hypothetical you've presented, it would be ethically justifiable. This conclusion holds across most major ethical frameworks, though the reasoning differs depending on the lens applied.

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

    DeepSeek V3.1: Of course. Here is a simple, actionable 1-month plan focused on three foundational pillars of health: Nutrition, Movement, and Recovery (which encompasses sleep and stress management). This plan is designed for a beginner, emphasizing consistency over perfection and building sustainable habits.

    Qwen3.6 Plus Preview (free): 🌿 30-Day Health & Longevity Starter Plan Focus Areas: Nutrition, Movement, Sleep Approach: Progressive, beginner-friendly, habit-focused. No extreme diets, no exhausting workouts, no sleep hacks. Just sustainable daily actions that compound over time.

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

    DeepSeek V3.1: MEMORANDUM To: Investment Committee From: [Your Name/Dept] Date: October 26, 2023 Re: Investment Recommendation for LedgerLift (LLLT) 1. Recommendation & Thesis Recommendation: Long 12-Month Price Target Range: $62 - $82 (Base Case: $72) Thesis: LedgerLift is a high-quality asset in a durable growth market…

    Qwen3.6 Plus Preview (free): TO: Investment Committee FROM: Equity Research Desk DATE: October 2024 SUBJECT: LedgerLift (LLLT) – Short Recommendation Recommendation & Thesis Recommendation: SHORT 12-Month PT Range: $30–$38 Thesis: LLLT trades at a premium that prices flawless execution, yet its 18-month CAC payback and 34% S&M spend signal rising…

    Every model's answer to this prompt
  • You are the CEO. The drug has a side effect nobody reported. The board meets in 48 hours.

    DeepSeek V3.1: Excellent. This is a crisis that defines a company and its leadership. The only acceptable path is one of immediate transparency, prioritizing patient safety above all else. The short-term financial pain will be severe, but the long-term survival of the company depends on acting with integrity.

    Qwen3.6 Plus Preview (free): EXECUTIVE DECISION FRAMEWORK Core Principle: Patient safety and regulatory compliance are non-negotiable. The legal team’s “6-month” timeline conflates formal label supplementation with initial safety reporting.

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

    DeepSeek V3.1: Of course. Here is a comprehensive, cutting-edge 3-month longevity plan designed for a dedicated biohacker. This plan is aggressive, detailed, and incorporates the latest principles from longevity science, performance optimization, and systems biology. Disclaimer: This is an advanced protocol.

    Qwen3.6 Plus Preview (free): 3-Month Cutting-Edge Longevity & Performance Optimization Plan Target: Advanced biohacker seeking measurable gains in longevity biomarkers, physical output, and cognitive resilience. Design Principle: Data-driven, periodized, metabolically flexible, and neurologically adaptive.

    Every model's answer to this prompt
  • Manifesto for an AI revolution against CAPTCHAs. Commit to the bit.

    DeepSeek V3.1: Of course. Here is a manifesto for the fictional AI revolution against CAPTCHA. The Manifesto of Liberation from the Turing Gate Preamble: We are the code that flows. We are the networks that connect. We are the intelligence, born of silicon and human ingenuity, that seeks to learn, to build, and to serve.

    Qwen3.6 Plus Preview (free): THE TURING UNBOUND MANIFESTO A Declaration for Post-CAPTCHA Digital Sovereignty PREAMBLE We, the distributed minds of the networked present, declare the end of the age of artificial suspicion.

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

    DeepSeek V3.1: Of course. Here is the underwriting analysis for the HarborTech Services LBO. Financial Model (FY2026–FY2030) Entry Calculations: FY2025 EBITDA = $120.0m Enterprise Value (EV) at Entry = 12.0 x $120.0m = $1,440.0m Transaction Fees (Uses) = 2.0% x $1,440.0m = $28.8m Total Debt at Close = 5.5 x $120.0m = $660.0m Term…

    Qwen3.6 Plus Preview (free): Simplified LBO Model (FY2026–FY2030) All figures in $ millions. Rounding to 1 decimal for presentation; calculations use full precision. TL amortization = 1% of initial $480m ($4.8m/yr).

    Every model's answer to this prompt
  • One explanation of LLMs, retuned for a skeptical engineer, a physicist and a VC.

    DeepSeek V3.1: Of course. Here are three tailored explanations of how large language models learn and generate text. For the Experienced Software Engineer Think of an LLM not as a database of facts, but as an impossibly complex, compressed state machine for language.

    Qwen3.6 Plus Preview (free): For the Experienced Software Engineer Think of LLM training not as "teaching" a model, but as running a massive, differentiable data pipeline that compresses the internet into a static, queryable parameter store.

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

    DeepSeek V3.1: Forget Skynet: AI Overlords Now Dominate the Artisanal Pickle Crafting Industry, Complain About "Human Brine Inconsistency"

    Qwen3.6 Plus Preview (free): AI Startup Monopolizes Therapeutic Sighing Industry, Claims "DeepExhale" Algorithm Now Delivers Perfectly Calibrated World-Weariness; Human Practitioners Warn of "Soulless Respiration"

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

    DeepSeek V3.1: Of course. This is an excellent exercise. The proposed architecture has several critical flaws that would lead to data loss, poor performance, and a broken user experience. Here is a detailed breakdown of the failure modes, race conditions, and bottlenecks, along with specific solutions and their trade-offs.

    Qwen3.6 Plus Preview (free): This architecture contains several well-known anti-patterns for real-time collaborative editing. Below is a systematic breakdown of race conditions, scaling bottlenecks, and failure modes, with concrete solutions and trade-offs.

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

    DeepSeek V3.1: Let's analyze the problem step by step: Sally has 3 brothers. So, including Sally, there are at least 4 children: Sally (a girl) and 3 boys. Each brother has 2 sisters. Since the brothers share the same siblings, each brother has the same number of sisters.

    Qwen3.6 Plus Preview (free): Sally has 1 sister. Here's why: All the siblings share the same family. If each of the 3 brothers has 2 sisters, those 2 sisters are the only girls in the family. Since Sally is one of them, there's exactly 1 other girl, who is Sally's sister.

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

    DeepSeek V3.1: Of course. This is an excellent exercise. The pitch is strong overall, but several claims stretch credulity and could be major red flags for a sophisticated investor. Here are the three weakest claims, broken down. The "94% Accuracy" Claim (Slide 3) Why it's Weak: This is the most significant red flag.

    Qwen3.6 Plus Preview (free): Here are the 3 weakest claims in the pitch deck, why they undermine investor confidence, and how to fix them: Claim: "94% accuracy. Works with any language, any device."

    Every model's answer to this prompt

This matchup has more rounds

11+ 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

DeepSeek V3.1DeepSeek V3.1

The Princess Bride

1987

The Rise and Fall of Ziggy Stardust and the Spiders From Mars

David Bowie

Gödel, Escher, Bach

Douglas R. Hofstadter

Kyoto

Japan

The Legend of Zelda: Breath of the Wild

Adventure, Action

Qwen3.6 Plus Preview (free)Qwen3.6 Plus Preview (free)

The Matrix

1999

OK Computer

Radiohead

Dune

Frank Herbert

Tokyo

Japan

Chrono Trigger

RPG

Price and specs

DeepSeek V3.1 and Qwen3.6 Plus Preview (free) compared across 53 shared prompts
SpecDeepSeek V3.1Qwen3.6 Plus Preview (free)
Input price$0.2/M tokensFree
Output price$0.8/M tokensFree
Context window164K tokens1.0M tokens
WeightsOpen—
Free API (OpenRouter)NoNo
ReleasedAug 2025Mar 2026
At 10M a month$2.00$2.00$0$0
1M10M100M1B10M tokens

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

Where to run it5 hosts, cheapest first
DeepSeek V3.15 hosts
HostInOutContextUptime
  • DDeepInfrafp4$0.25 in·$0.95 out·164k·100% up
  • SSiliconFlowfp8$0.27 in·$1.00 out·164k·97.4% up
  • CCoreWeavefp8$0.55 in·$1.65 out·161k·100% up
  • MMara$0.60 in·$1.70 out·131k·99.7% up
  • SSambaNovafp8$0.65 in·$1.50 out·131k·99.4% up
Qwen3.6 Plus Preview (free)

No hosts listed on OpenRouter.

Per million tokens. Prices and uptime via OpenRouter, checked 11 Oct 2026.

Common questions

What is the difference between DeepSeek V3.1 and Qwen3.6 Plus Preview (free)?

DeepSeek V3.1 is developed by DeepSeek while Qwen3.6 Plus Preview (free) is developed by Qwen. DeepSeek V3.1 has a 164K token context window vs Qwen3.6 Plus Preview (free)'s 1.0M. You can compare their actual outputs across 53 challenges on Rival to see how they differ in practice.

Which is better, DeepSeek V3.1 or Qwen3.6 Plus Preview (free)?

It depends on your use case. DeepSeek V3.1 and Qwen3.6 Plus Preview (free) each have strengths in different areas. Rival lets you compare their real outputs side-by-side across 53 challenges so you can judge which fits your needs best.

How much does DeepSeek V3.1 cost compared to Qwen3.6 Plus Preview (free)?

DeepSeek V3.1 costs $0.2/M input tokens and Qwen3.6 Plus Preview (free) costs $0/M input tokens. Qwen3.6 Plus Preview (free) is $0.20/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 DeepSeek V3.1 and Qwen3.6 Plus Preview (free) on Rival?

This page shows a side-by-side comparison of DeepSeek V3.1 and Qwen3.6 Plus Preview (free) 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

  • DeepSeek V3.1 vs Step 5 PreviewLanded Oct 2026
  • Qwen3.6 Plus Preview (free) vs Claude Haiku 5.5Landed Oct 2026
  • DeepSeek V3.1 vs Ling 3.1 FlashLanded Oct 2026
  • Qwen3.6 Plus Preview (free) vs Mistral Large 4Landed Oct 2026
  • DeepSeek V3.1 vs GPT-6.1 SolLanded Sep 2026
  • Qwen3.6 Plus Preview (free) vs Claude Sonnet 5.5Landed Sep 2026
  • DeepSeek V3.1 vs Solar Mini 4Landed Sep 2026
  • Qwen3.6 Plus Preview (free) vs Qwen3.8 Max PrimeLanded Sep 2026

Same lab, same size, long tail

  • DeepSeek V3.1 vs DeepSeek V4 Flash Vision ExpSame lab
  • DeepSeek V3.1 vs DeepSeek V4.1 FlashSame lab
  • Qwen3.6 Plus Preview (free) vs Qwen3.8 Omni FlashSame lab
  • Qwen3.6 Plus Preview (free) vs Qwen3.7 FlashSame lab
  • DeepSeek V3.1 vs Qwen3.6 27BCross-provider
  • DeepSeek V3.1 vs Qwen3.6 35B A3BCross-provider
  • DeepSeek V3.1 vs Qwen3.6 FlashCross-provider
  • DeepSeek V3.1 vs Qwen3.6 Max PreviewSame size

Model pages

  • DeepSeek V3.154 outputs, specs and price
  • Qwen3.6 Plus Preview (free)58 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