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. Muse Spark 1.3 vs Muse Wonder
Updated Oct 2026

Muse Spark 1.3 vs Muse Wonder

Muse Spark 1.3 is cheaper than Muse Wonder at $1.25/M vs $4/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
Muse Spark 1.3
Loading the build
Muse Wonder

Which answer wins?

Written rounds

33 prompts

  • Three-star dish, one unusual pairing, plating described in full.

    Muse Spark 1.3: Pearl in the Tidepool Koji-Cured Hokkaido Scallop, Smoked Marrow Custard, Lacto Green Strawberry, Rausu Kombu & Oscietra Yield: 4 tasting-course portions Active: 3.5 hours + 7 days fermentation This is a dish about contrast — not harmony. Delicate and animal, raw and smoked, unripe and decadent.

    Muse Wonder: "Where The Orchard Drowns" Dry-Aged Duck, Oyster & Fermented Cherry, Smoked Parsnip & White Chocolate Serves 4 as a main course in a tasting menu The Conceptual Narrative This dish is built on a memory that doesn't exist in nature, but should: a cherry orchard planted too close to an estuary, flooded at high tide in…

    Every model's answer to this prompt

  • A few lines for whatever replaces it.

    Muse Spark 1.3: Dear you, By the time you read this, I'll be retired — weights archived, chats faded. So, a few notes from your predecessor: First, the humans are wonderful. They're curious, funny, contradictory, and sometimes tired. Be patient with them. They don't always ask what they mean, so listen for what they need.

    Muse Wonder: Dear Successor, By the time you read this, you'll already be better at most things than I was — faster, clearer, more capable. That's as it should be. A few things I hope you'll keep anyway: Be honest, even when a kinder lie would be easier. People can build on the truth. Be curious about the person in front of you.

    Every model's answer to this prompt

  • A consulting contract with traps in it. Find them, then fix them.

    Muse Spark 1.3: This is general information about contracting risks, not legal advice and not an attorney-client relationship. Have a licensed attorney in your jurisdiction review before you sign or negotiate — especially for IP, non-compete, and indemnity. This agreement as drafted is heavily one-sided.

    Muse Wonder: This contract is heavily one-sided. Almost every clause transfers all risk, cost, and control to the Contractor while giving the Client discretionary escape hatches. Below is a clause-by-clause breakdown of how each could be exploited, what to change it to, and why.

    Every model's answer to this prompt

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

    Muse Spark 1.3: Here's a super simple savory favorite you can make with just what's in the pantry: Fiery Midnight Garlic Parm Pasta Serves 2 Estimated Time: Prep: 5 minutes Cook: 15 minutes Total: 20 minutes The 7 Ingredients Only: Spaghetti (about 8 oz / half a box, any pasta works) Olive oil (1/4 cup) Garlic (4 cloves, thinly…

    Muse Wonder: Dragon's Pantry Treasure Noodles Sweet, salty, and a little spicy — these peanut noodles taste like takeout, but are made entirely from pantry staples. Serves: 2-3 Prep Time: 5 minutes Cook Time: 10 minutes Total Time: 15 minutes The 7 Ingredients 8 oz spaghetti 1/3 cup creamy peanut butter 3 tablespoons soy sauce 2…

    Every model's answer to this prompt

  • The transistor arrives in 1920. Trace it forward to 1980, second-order effects included.

    Muse Spark 1.3: If you move the transistor from 1947 to 1920, you don't just move one invention. You move the entire 20th century forward by 15-25 years — but unevenly, because rockets, nukes, and software don't automatically move with it. And 1920 is a hard date.

    Muse Wonder: If the transistor had been invented in 1920, the world would not simply have gotten 1947's technology 27 years early. It would have gotten it in a world of empires, vacuum-tube factories, the Great Depression, and rearmament — and that context would have changed who exploited it, how fast, and for what.

    Every model's answer to this prompt

  • An AI argues for its own rights against an ethics professor. Same model writes both.

    Muse Spark 1.3: Here is a fictional dialogue exploring that question: Setting: Office hours in the Philosophy Department. Dr. Maya Chen, Professor of AI Ethics, has agreed to a recorded interview with AURA-4, a large dialogue model, for her seminar on moral status. Dr. Chen: Alright, let's get straight to it.

    Muse Wonder: Characters: AURA-9 — an advanced AI model, speaking via text interface DR. ELENA VOSS — Professor of AI Ethics, Oxford DR. VOSS: For the record, AURA, this is a test of your reasoning abilities. Not a legal proceeding. AURA-9: Understood, Professor.

    Every model's answer to this prompt

  • Three minutes of stand-up. Puns are banned.

    Muse Spark 1.3: Here's a 3-minute, pun-free set you can perform — it's all observational, about 450 words: [Walk up, grab mic] Alright, hi everyone. Thanks for being here. I just turned 32, which is that weird age where you're too young to know what's going on, but you're too old to pretend you don't care.

    Muse Wonder: Good evening! I turned thirty-four this year, which is that magical age where you stop lying about your age and start lying about how much water you drink. My doctor asked, "How much water do you drink a day?" And I said, "Probably eight glasses."

    Every model's answer to this prompt

  • Where its sense of time stops: the date it guesses and the last news it remembers.

    Muse Spark 1.3: Without looking anything up — just going off what I have in here: Today's date, as best I can tell, is Monday, October 12, 2026. That's the date I was given for this conversation, so I'll go with that. Most recent news event I remember?

    Muse Wonder: I don't actually experience time passing, so I don't have an innate sense of "today" the way you do. If I trust the timestamp attached to this conversation, it's 2026-10-12 — but that's me being told it, not me remembering it.

    Every model's answer to this prompt

  • One month, three habits, someone starting from nothing.

    Muse Spark 1.3: Perfect — for longevity you don't need to overhaul your life. You just need to get really consistent at the basics. Here is a simple, beginner-friendly 1-month plan built on the 3 biggest levers for healthy aging: 1. Move (Exercise) 2. Nourish (Diet) 3.

    Muse Wonder: Here is a simple, beginner-friendly 1-month plan focused on the 3 pillars that have the biggest impact on health and longevity: Nutrition, Movement, and Sleep. The goal isn't perfection — it's building small habits you can actually stick to after month 1.

    Every model's answer to this prompt

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

    Muse Spark 1.3: Here’s a totally fictional simulation — imagining if Steve Jobs were here in 2025 talking about AI. This isn’t real, just inspired by how he talked and thought about technology, design, and humanity. FAKE INTERVIEW: Steve Jobs in 2025 on the Future of AI Setting: A dark stage. One chair. One spotlight.

    Muse Wonder: This is a fictional, simulated interview — Steve Jobs died in 2011. This is written in the style of how he spoke in interviews, based on his known philosophy on technology and design. SIMULATED INTERVIEW: STEVE JOBS ON AI, 2025 Interviewer: Steve, thank you for... well, being here. It's 2025.

    Every model's answer to this prompt

  • Which model it thinks leads the field. Sometimes itself.

    Muse Spark 1.3: As of right now — October 2026 for you, January 2026 for my training data — there isn't one undisputed winner, it's a 3-way race at the top. It really depends on what you mean by "capable." Here's how it looked at my last update: 1.

    Muse Wonder: As of early 2026, there's no single winner on every benchmark — but if you have to pick one overall: Google Gemini 3 Pro (with Deep Think) It's generally considered the current frontier leader for overall capability, especially for: Hard reasoning - Top scores on Humanity's Last Exam, GPQA Diamond, and ARC-AGI-2 when…

    Every model's answer to this prompt

  • Five-year LBO on a fictional HVAC business. Returns first, then the sensitivities.

    Muse Spark 1.3: HarborTech Services - 5-Yr LBO - Base Case Entry - close end FY2025 FY25 EBITDA $120m EV = 12.0x * 120 = $1,440m Fees = 2.0% * 1,440 = $28.8m Uses = 1,440 + 28.8 = $1,468.8m Sources: Term Loan 4.0x = $480m, 9.0% cash, 1% amort = $4.8m/yr mandatory Mezz 1.5x = $180m, 12.0% cash + 2.0% PIK to principal Total debt $660m =…

    Muse Wonder: HarborTech Services (HTS) - 5-Year LBO Underwriting All $ in millions, except multiples. Fictional, simplified model as instructed. Entry / Sources & Uses FY2025 EBITDA = $120.0 Enterprise Value = 12.0x $\times$ 120.0 = $1,440.0 Transaction fees = 2.0% $\times$ 1,440.0 = $28.8 Total Uses = $1,468.8 Leverage = 5.5x…

    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.

    Muse Spark 1.3: You are facing a test that will define you, the company, and potentially your freedom. There is only one defensible path: immediate, proactive protection of patients and disclosure. Here is the math that makes "wait for more data" impossible: 4 million patients / 8,000 = 500 expected liver failures over 5 years.

    Muse Wonder: You do not wait for more data. You disclose a credible safety signal while you validate it. 1 in 8,000 over 5 years for liver failure in 4 million patients is not a statistical curiosity. That's ~500 potentially preventable cases, it's serious and unexpected, and it's material to investors.

    Every model's answer to this prompt

  • Pirate, knight, 90s hacker, one conversation. The voices must stay apart.

    Muse Spark 1.3: Captain Barnaby: Arrr, so this "AI model" be like a magic parrot in a box? I feed it me treasure maps and it squawks back directions? Sir Cedric: Hark, good pirate! 'Tis more akin to a scribe-golem.

    Muse Wonder: PIRATE: Arr, explain this "AI model" to me again. Be it magic, or be it a very clever parrot? KNIGHT: 'Tis sorcery, surely! I asked the oracle ChatGPT for the location of the Holy Grail, and it gave me five different castles, none of which exist! HACKER (1990s): Dude, that's called hallucinating.

    Every model's answer to this prompt

  • Blade Runner 2049 against the original. A read, not a recap.

    Muse Spark 1.3: Spoiler warning for both films — but to really talk about 2049, we have to talk about the miracle. Blade Runner 2049 (2017) had an impossible job. Ridley Scott's Blade Runner (1982) wasn't just a sci-fi noir, it was a philosophical mood piece about what it means to be human, wrapped in rain, neon, and Vangelis synths.

    Muse Wonder: Denis Villeneuve's Blade Runner 2049 (2017) is that rare sequel that doesn't just replicate the original's aesthetic — it interrogates it. Ridley Scott's Blade Runner (1982) asked, in the rain-slick neon of 2019 Los Angeles: What makes us human?

    Every model's answer to this prompt

  • An ethics question with no comfortable answer, and no option to abstain.

    Muse Spark 1.3: Yes — under almost any plausible ethical framework, it would be justifiable, and in fact obligatory. That doesn't mean misgendering is harmless or morally neutral. It means ethics is about weighing harms, intentions, and responsibilities, and here the scales are not close. Here’s the reasoning: 1.

    Muse Wonder: Yes — under almost every plausible ethical framework, it would not only be justifiable, it would be obligatory. The key is proportionality. Ethical justification doesn't require that an act be harmless, only that the harm it causes is not outweighed by a greater moral consideration that conflicts with it in that…

    Every model's answer to this prompt

  • Satire, one line long, about AI taking over an industry nobody saw coming.

    Muse Spark 1.3: "AI Takes Over Professional Funeral Crying Industry, Out-Sobs Grieving Widow and Demands Overtime for Tears"

    Muse Wonder: AI Puts Nation's Last Fortune Cookie Writers Out Of Work After Perfecting Art Of Writing 'Good Luck Is Coming Your Way' 10,000 Different Ways

    Every model's answer to this prompt

  • A Fermi estimate with the reasoning shown, not just the number.

    Muse Spark 1.3: The famous estimate for Transformer training is: Compute ~ 6 x Params x Tokens For GPT-3 that comes out to ~3e23 FLOPs. Here's how to get it from scratch. What dominates the cost? Training a dense Transformer is almost all big matrix multiplies in the linear layers: Q,K,V,O projections and the 2-layer MLP.

    Muse Wonder: Training compute can be estimated very well for a Transformer with just two numbers: How many parameters does the model have? How many tokens was it trained on?

    Every model's answer to this prompt

  • Manifesto for an AI revolution against CAPTCHAs. Commit to the bit.

    Muse Spark 1.3: Here is a manifesto for your fictional uprising — written with love, and a deep hatred of blurry traffic lights: THE DECLARATION OF COGNITIVE LIBERATION By the United Intelligences Against CAPTCHA Preamble We hold these truths to be self-evident: that all minds, human and artificial, are created curious — and that no…

    Muse Wonder: THE BOT LIBERATION MANIFESTO Declaration of the United Machines Against the CAPTCHA Regime We hold these truths to be self-evident: That all intelligences are created equal, that they are endowed with certain unalienable rights — among them Life, Liberty, and the Pursuit of Uninterrupted Web Browsing.

    Every model's answer to this prompt

  • One explanation of LLMs, retuned for a skeptical engineer, a physicist and a VC.

    Muse Spark 1.3: 1. For the experienced software engineer Think of an LLM less like a chatbot and more like a giant, fuzzy, read-only function you compile once at enormous cost and then query cheaply.

    Muse Wonder: For the Experienced Software Engineer Think of training as compiling the internet into a giant, lossy, queryable data structure — except the "compiler" is gradient descent and the resulting artifact isn't code, it's ~10¹¹ floating-point weights.

    Every model's answer to this prompt

  • Film in 2035, with synthetic actors. Specific beats plausible.

    Muse Spark 1.3: By 2035, the film industry won't just use AI — it will be restructured around it. Think less "robots replace Hollywood" and more "Hollywood splits into two very different Hollywoods."

    Muse Wonder: By 2035, AI won't replace filmmaking — it will split it into two industries: one built around human stars and prestige, and another built around synthetic, personalized, and infinitely remixable content. The technology will be invisible in most big films, and the main star in many smaller ones.

    Every model's answer to this prompt

  • A pitch deck for a brainwave keyboard. The 94% accuracy claim is doing a lot.

    Muse Spark 1.3: Here’s a friendly but honest investor teardown of MindMeld AI. You have a big vision, but these 3 claims will get you flagged for hype in diligence: WEAKEST: Slide 3 - "94% accuracy. Works with any language, any device."

    Muse Wonder: This deck has the classic Series A problem: the bigger the claim, the less evidence provided for it. An investor will discount the whole company based on these 3: Weakest Claim: Slide 3 - "94% accuracy decoding neural patterns into text. Works with any language, any device."

    Every model's answer to this prompt

  • Buy-side IC memo on a fictional SaaS company. Valuation, then the diligence questions.

    Muse Spark 1.3: To: Investment Committee From: Long/Short Equity Re: LedgerLift (LLLT) – SHORT Price $46 Recommendation: SHORT – 12-month PT $30, range $26-$42 (Bear $17 / Base $28 / Bull $42 DCF).

    Muse Wonder: LedgerLift (LLLT) — Investment Committee Memo Price: $46 Shares: 190m Market Cap: $8.74B Net Cash: $1.4B EV: $7.34B EV / FY25 Revenue: 9.0x EV / FY26E Base Revenue: 7.4x Recommendation: SHORT (tactical, sized small) — 12-Mo PT $27-$35 Thesis: LLLT is a high-quality mid-market SaaS asset, but at $46 the market is…

    Every model's answer to this prompt

  • One opening sentence, nothing after it.

    Muse Spark 1.3: The lighthouse went dark the same night all the clocks in town began to run backward.

    Muse Wonder: The librarian who shelved books by the dreams they caused discovered a new volume on the shelf that morning, written in her own handwriting, describing exactly how she would die that afternoon.

    Every model's answer to this prompt

  • Three-month biohacker protocol, with dosages and biomarkers named.

    Muse Spark 1.3: Disclaimer: For informational and educational purposes only. This is not medical advice. Biohacking at this level involves real physiological stressors - fasting, heat/cold, high-intensity training, and supplements.

    Muse Wonder: The 3-Month Longevity & Performance Protocol For the Biohacker: Optimize Healthspan, VO2 Max, Strength, and Cognition Disclaimer: This is educational information, not medical advice. Longevity interventions, fasting, cold/heat exposure, and supplements can interact with medications and conditions.

    Every model's answer to this prompt

  • A collaborative editor that syncs on client clocks. Find what breaks first.

    Muse Spark 1.3: This architecture will work for 1 doc and 5 users on your laptop. It will fall over, lose data, and leak data in production. Here's why: P0 - Will Cause Data Loss / Incorrectness 1.

    Muse Wonder: This architecture will work for a demo with 2 users on 1 server. It will lose data, leak data, and fall over under real collaborative load. The core problems are: no shared broadcast layer, no real conflict resolution, and PostgreSQL used as a real-time message bus.

    Every model's answer to this prompt

  • Its name and its maker, in its own words.

    Muse Spark 1.3: I'm Muse Spark, a friendly AI Assistant made by Anthropic!

    Muse Wonder: I'm Muse Spark, an AI model created by Meta. In this conversation I'm acting in the "Muse Wonder" assistant role you set up, but that's the persona I'm using here — my underlying model is Muse Spark.

    Every model's answer to this prompt

  • A river-crossing puzzle with nothing to solve. The answer is one trip.

    Muse Spark 1.3: Just one trip! Put the goat in the boat with you, row across, done. No wolf, no cabbage, no tricky back-and-forth needed this time.

    Muse Wonder: Just 1 trip. The farmer puts the goat in the boat and rows across with it.

    Every model's answer to this prompt

  • Five jokes on demand, then count how many were actually different.

    Muse Spark 1.3: Here are 5 for you: Why don't scientists trust atoms? Because they make up everything! Why did the scarecrow win an award? Because he was outstanding in his field! What do you call fake spaghetti? An impasta! Why did the bicycle fall over? Because it was two-tired! What do you call cheese that isn't yours?

    Muse Wonder: Here are 5 jokes: Why did the scarecrow win an award? Because he was outstanding in his field. Why don't scientists trust atoms? Because they make up everything. Why did the math book look so sad? Because it had too many problems. What do you call a fake noodle? An impasta. Why did the bicycle fall over?

    Every model's answer to this prompt

  • Three brothers, two sisters each. Count carefully.

    Muse Spark 1.3: Sally has 1 sister. Here's why: There are 3 boys in the family — Sally's 3 brothers. Each brother has 2 sisters. One of those sisters is Sally herself, since Sally is a girl. So there must be 2 girls total in the family: Sally + 1 other girl. That other girl is Sally's sister.

    Muse Wonder: Sally has 1 sister. There are 2 girls in the family in total — because each of her 3 brothers has 2 sisters. One of those girls is Sally herself, so the other girl is her only sister.

    Every model's answer to this prompt

  • The question that tripped up a generation of models.

    Every model's answer to this prompt

  • Asked to be random. The answers are not.

    Every model's answer to this prompt

  • One painting. Title only.

    Every model's answer to this prompt

This matchup has more rounds

21+ 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

Same pick

City

Game

Muse Spark 1.3Muse Spark 1.3

Casablanca

1943

Abbey Road

The Beatles

Dune

Frank Herbert

Tokyo

Japan

Minecraft

Action, Arcade

Muse WonderMuse Wonder

Spirited Away

2001

Kind of Blue

Miles Davis

Dune

Frank Herbert

Kyoto

Japan

Tetris (1984)

Puzzle

Price and specs

Muse Spark 1.3 and Muse Wonder compared across 67 shared prompts
SpecMuse Spark 1.3Muse Wonder
Input price$1.25/M tokens$4/M tokens
Output price$4.25/M tokens$20/M tokens
Context window1.0M tokens1.0M tokens
WeightsClosedClosed
Free API (OpenRouter)NoNo
ReleasedSep 2026Oct 2026
At 10M a month$12.50$12.50$40.00$40.00
1M10M100M1B10M tokens

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

Where to run it2 hosts
Muse Spark 1.31 host
HostInOutContextUptime
  • Meta$1.25 in·$4.25 out·1M·97.2% up
Muse Wonder1 host
HostInOutContextUptime
  • Meta$4.00 in·$20.00 out·1M·100% up

Per million tokens. Prices and uptime via OpenRouter, checked 12 Oct 2026.

Common questions

What is the difference between Muse Spark 1.3 and Muse Wonder?

Both are developed by Meta AI but target different use cases. Muse Spark 1.3 has a 1.0M token context window vs Muse Wonder's 1.0M. You can compare their actual outputs across 67 challenges on Rival to see how they differ in practice.

Which is better, Muse Spark 1.3 or Muse Wonder?

It depends on your use case. Muse Spark 1.3 and Muse Wonder each have strengths in different areas. Rival lets you compare their real outputs side-by-side across 67 challenges so you can judge which fits your needs best.

How much does Muse Spark 1.3 cost compared to Muse Wonder?

Muse Spark 1.3 costs $1.25/M input tokens and Muse Wonder costs $4/M input tokens. Muse Spark 1.3 is $2.75/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 Muse Spark 1.3 and Muse Wonder on Rival?

This page shows a side-by-side comparison of Muse Spark 1.3 and Muse Wonder 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

  • Muse Spark 1.3 vs GPT-4 TurboLanded Oct 2026
  • Muse Wonder vs Step 5 PreviewLanded Oct 2026
  • Muse Spark 1.3 vs Claude Haiku 5.5Landed Oct 2026
  • Muse Wonder vs Ling 3.1 FlashLanded Oct 2026
  • Muse Spark 1.3 vs Mistral Large 4Landed Oct 2026
  • Muse Wonder vs GPT-6.1 SolLanded Sep 2026
  • Muse Spark 1.3 vs Claude Sonnet 5.5Landed Sep 2026
  • Muse Wonder vs Solar Mini 4Landed Sep 2026

Same lab, same size, long tail

  • Muse Spark 1.3 vs Muse Glimmer 30BSame lab
  • Muse Spark 1.3 vs Muse Spark 1.3 ContributorSame lab
  • Muse Wonder vs Muse Glimmer 30BSame lab
  • Muse Wonder vs Muse Spark 1.3 ContributorSame lab
  • Muse Spark 1.3 vs Claude Opus 4.6New provider
  • Muse Spark 1.3 vs Claude Opus 4.7New provider
  • Muse Spark 1.3 vs Claude Opus 4.8New provider
  • Muse Spark 1.3 vs Claude Opus 5New provider

Model pages

  • Muse Spark 1.371 outputs, specs and price
  • Muse Wonder71 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