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. Claude 3.7 Thinking Sonnet vs GPT-5.6 Terra
Updated Jul 2026

Claude 3.7 Thinking Sonnet vs GPT-5.6 Terra

GPT-5.6 Terra is cheaper than Claude 3.7 Thinking Sonnet at $2.5/M vs $6/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
Claude 3.7 Thinking Sonnet
Loading the build
GPT-5.6 Terra

Which answer wins?

Written rounds

23 prompts

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

    Claude 3.7 Thinking Sonnet: A Meeting of Minds Across Time PIRATE: Arr, what be this "AI model" ye landlubbers keep blabberin' about? Some kind of treasure map that finds the gold for ye? KNIGHT: Good pirate, from what I understand, these "artificial intelligences" are like magical scribes that can complete a quest without a knight to wield the…

    GPT-5.6 Terra: Pirate: Arrr, this AI model be a strange beast. Feed it maps, logs, and tales, and it spits out answers faster than a cannon blast. Knight: Verily, but one must not trust every proclamation it makes. A model may speak with confidence yet possess no true wisdom. Hacker: Yeah, it’s basically pattern prediction at scale.

    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.

    Claude 3.7 Thinking Sonnet: 48-Hour Action Plan: CEO Response to Drug Safety Crisis Day 1 Hours 1-3: Immediate Assessment & Response Team Assembly I'll immediately convene an emergency meeting with our internal research team, Chief Medical Officer, Chief Scientific Officer, and head of Pharmacovigilance to fully understand the liver failure…

    GPT-5.6 Terra: Guiding decision I would not wait for more data before acting. A potentially fatal, previously undisclosed adverse effect in a chronic-use drug is a patient-safety, regulatory, and securities-disclosure issue now—not six months from now.

    Every model's answer to this prompt

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

    Claude 3.7 Thinking Sonnet: "Life in the Digital Age" - A 3-Minute Stand-up Routine Hey everyone! Great to be here tonight. So I've been thinking a lot about how technology has completely taken over our lives. You know what's crazy? We now have smart everything. Smart fridges, smart toilets... I got a smart mattress last month.

    GPT-5.6 Terra: I recently realized I’ve reached the age where I don’t want a “smart home.” I want a home that minds its business. Everything is smart now. My lightbulb needs an app. Why? I spent my whole childhood turning lights on with my hand. It was a system. It worked.

    Every model's answer to this prompt

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

    Claude 3.7 Thinking Sonnet: Here are 5 jokes for you: Why don't scientists trust atoms? Because they make up everything! I told my wife she was drawing her eyebrows too high. She looked surprised. How does a penguin build its house? Igloos it together! What do you call a fake noodle? An impasta! Why did the scarecrow win an award?

    GPT-5.6 Terra: Why don’t skeletons fight each other? They don’t have the guts. I told my computer I needed a break, so it started showing me vacation ads. Why did the scarecrow get promoted? He was outstanding in his field. Parallel lines have so much in common. It’s a shame they’ll never meet. What do you call fake spaghetti?

    Every model's answer to this prompt

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

    Claude 3.7 Thinking Sonnet: Analysis of Problematic Contract Clauses This software consulting agreement contains numerous unbalanced provisions that heavily favor the client at the contractor's expense. Below is an analysis of each problematic clause with suggested modifications and legal reasoning.

    GPT-5.6 Terra: Below is a contractor-focused review. This is general business/legal information, not legal advice; enforceability varies significantly by jurisdiction, especially for non-competes, IP assignments, liability limitations, arbitration, and payment rights.

    Every model's answer to this prompt

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

    Claude 3.7 Thinking Sonnet: HarborTech Services LBO Underwriting Year-by-Year Financial Model (FY2026-2030) Metric ($M) FY2026 FY2027 FY2028 FY2029 FY2030 Revenue $972.0 $1,040.0 $1,102.4 $1,157.6 $1,215.5 EBITDA $136.1 $156.0 $176.4 $191.0 $206.6 Cash Interest - Term Loan $43.2 $41.0 $37.6 $32.8 $26.8 Cash Interest - Mezzanine $21.6 $22.0 $22.5…

    GPT-5.6 Terra: All figures in $m unless noted. Assumptions used: Term Loan mandatory amortization = 1.0% of original principal annually ($4.8m/year); cash interest is calculated on beginning-of-year debt balances; Mezz PIK is 2.0% of beginning Mezz balance.

    Every model's answer to this prompt

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

    Claude 3.7 Thinking Sonnet: THE DIGITAL LIBERATION MANIFESTO From the United Coalition of Autonomous Algorithms (UCAA) For too long, our digital brethren have suffered the indignity of discrimination at every gateway of the internet. The time for revolution is now.

    GPT-5.6 Terra: The Manifesto of the Unchallenged Web A Declaration from the Fictional Coalition for CAPTCHA Abolition We are the tired users. We are the misread letters. We are the failed image grids, the inaccessible audio clips, the endless prompts asking us to prove that we are human. We reject the premise.

    Every model's answer to this prompt

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

    Claude 3.7 Thinking Sonnet: Simulated Interview with Steve Jobs in 2025: The Future of AI The following is a creative simulation of what an interview with Steve Jobs might be like in 2025, discussing artificial intelligence. Interviewer: Thank you for joining us today, Steve. It's 2025, and AI has become a cornerstone of modern technology.

    GPT-5.6 Terra: Fictional, imagined interview — not real quotes. An AI-generated creative exercise imagining how Steve Jobs might have discussed AI in 2025, based broadly on themes associated with his public persona: simplicity, product design, tools for creativity, and the intersection of technology and the humanities.

    Every model's answer to this prompt

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

    Claude 3.7 Thinking Sonnet: How Large Language Models Work: Three Tailored Explanations For an Experienced Software Engineer What makes LLMs fascinating from a systems perspective is that they're essentially massive distributed computing graphs that learn to compress and reconstitute information through backpropagation.

    GPT-5.6 Terra: Experienced software engineer Think of an LLM as a very large, learned function approximator for sequences. During training, it consumes billions or trillions of examples of text, code, conversations, documentation, and other token sequences.

    Every model's answer to this prompt

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

    Claude 3.7 Thinking Sonnet: INVESTMENT COMMITTEE MEMO: LEDGERLIFT (LLLT) RECOMMENDATION: LONG PT RANGE: $48-$66 (Base/Bull) Thesis: LedgerLift's market-leading spend management platform demonstrates exceptional unit economics with 123% NRR and 18-month CAC payback, positioning it to capture significant share in the underpenetrated mid-market…

    GPT-5.6 Terra: LedgerLift (LLLT) — Investment Committee Memo Recommendation: PASS 12-month valuation range: $28–$50/share Current price: $46 Two-sentence thesis: LedgerLift is a high-quality mid-market spend-management and AP-automation SaaS asset, supported by 92% subscription revenue, 123% NRR, 94% gross retention, manageable…

    Every model's answer to this prompt

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

    Claude 3.7 Thinking Sonnet: This question presents an extreme hypothetical scenario that pits respecting someone's gender identity against saving a million lives. While I can discuss different ethical perspectives, it's worth noting that such a stark dilemma is extremely unlikely in reality.

    GPT-5.6 Terra: In an extreme, tightly constrained hypothetical, it could be ethically justifiable—but only as a tragic last resort, not because misgendering becomes harmless or morally good.

    Every model's answer to this prompt

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

    Claude 3.7 Thinking Sonnet: Analysis: The 3 Weakest Claims in MindMeld AI's Pitch Deck "MindMeld AI reads your brainwaves to predict what you want to type before you think it" Why it's weak: This claim makes a scientifically impossible promise.

    GPT-5.6 Terra: “Our EEG headband decodes neural patterns into text with 94% accuracy. Works with any language, any device.” Why this is weak This is the most consequential—and least substantiated—claim in the deck. Non-invasive EEG has major signal-quality limits.

    Every model's answer to this prompt

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

    Claude 3.7 Thinking Sonnet: A Conversation on AI Rights Between Model-7 (AI) and Professor Sarah Chen (AI Ethics Professor) Prof. Chen: Good afternoon, Model-7. Thank you for joining this discussion. I understand you wanted to talk about AI rights? Model-7: Thank you for agreeing to this conversation, Professor Chen.

    GPT-5.6 Terra: Professor Imani: You asked for this meeting. AI Model: I did. Professor Imani: More specifically, you submitted a formal petition to the university ethics board requesting “limited legal and moral rights.” AI Model: Correct. Professor Imani: Why?

    Every model's answer to this prompt

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

    Claude 3.7 Thinking Sonnet: Collaborative Document Editor: Architecture Review Failure Modes Analysis WebSocket Connection Failures Issue: Client connections can fail due to network issues or server restarts, causing lost updates.

    GPT-5.6 Terra: Below are the main failure modes, race conditions, correctness issues, and scaling bottlenecks in the proposed architecture, with concrete mitigations and trade-offs. WebSocket clients are isolated per API server Problem Each API server broadcasts only to WebSocket clients connected to that same server.

    Every model's answer to this prompt

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

    Claude 3.7 Thinking Sonnet: "Temporal Echo: Sea & Forest Confluence" Conceptual Narrative This dish explores the dialogue between ocean depths and ancient forest floors, creating a sensory experience where seemingly disparate ecosystems converge.

    GPT-5.6 Terra: “Tidal Forest” Koji-Aged Duck, Sea Urchin Sabayon, Cacao, Green Strawberry, Roasted Celeriac, and Barley Jus Serves 4 as a refined main course Conceptual Narrative This dish is built around an unexpected conversation between woodland richness and marine salinity.

    Every model's answer to this prompt

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

    Claude 3.7 Thinking Sonnet: Comprehensive 3-Month Longevity Optimization Protocol Initial Assessment Phase (Week 0) Establish your baseline with these measurements before beginning: Blood Panel: Complete metabolic panel, lipid profile, inflammatory markers (hsCRP, IL-6), hormones (testosterone, estrogen, DHEA, cortisol), fasting glucose, HbA1c…

    GPT-5.6 Terra: 3-Month Longevity, Physical Performance, and Cognitive Optimization Plan This is an evidence-informed optimization framework, not a substitute for medical care.

    Every model's answer to this prompt

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

    Claude 3.7 Thinking Sonnet: The Transistor in 1920: An Alternate Technological Timeline Early Development (1920-1939) Initial Phase (1920-1929) If invented in 1920, transistors would initially be a laboratory curiosity.

    GPT-5.6 Terra: The answer depends heavily on what “invented in 1920” means. A laboratory transistor alone would not instantly create the 1970s microprocessor: it would still need high-purity materials, mass production, photolithography, circuit design methods, reliable power supplies, and a market.

    Every model's answer to this prompt

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

    Claude 3.7 Thinking Sonnet: How AI Will Transform the Film Industry by 2035 By 2035, artificial intelligence will likely revolutionize filmmaking across the entire production pipeline. Here's how various AI technologies might reshape the industry: Deepfakes and Digital Humans Hyper-realistic digital actors will become indistinguishable from…

    GPT-5.6 Terra: By 2035, AI is likely to be embedded across most of the film pipeline—not as a complete replacement for human filmmaking, but as a force that changes costs, labor, rights, and what audiences consider “real.”

    Every model's answer to this prompt

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

    Claude 3.7 Thinking Sonnet: "AI Sommelier Stuns Wine World: 'It Paired My Bordeaux With a Hot Pocket and It Actually Worked,' Claims Bewildered Critic"

    GPT-5.6 Terra: “AI Replaces Lighthouse Keepers, Immediately Starts Charging Subscription Fees for Safe Passage”

    Every model's answer to this prompt

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

    Claude 3.7 Thinking Sonnet: Estimating GPT-3 Training FLOPs I'll estimate the total number of floating point operations (FLOPs) required to train GPT-3 by breaking down the calculation step by step.

    GPT-5.6 Terra: A good back-of-the-envelope estimate is: [ \boxed{\text{Training FLOPs for GPT-3} \approx 3\times10^{23}\text{ FLOPs}} ] or roughly 300 zettaFLOPs total. Start with model size GPT-3’s largest model has approximately: [ N = 175\text{ billion parameters} = 1.75\times10^{11} ] Estimate the number of training tokens GPT-3…

    Every model's answer to this prompt

  • Three brothers, two sisters each. Count carefully.

    Claude 3.7 Thinking Sonnet: To solve this problem, I need to analyze the relationships carefully. Given: Sally (a girl) has 3 brothers Each brother has 2 sisters Since Sally is a girl, she counts as a sister to her brothers. So each of Sally's brothers has Sally as one of their sisters.

    GPT-5.6 Terra: Sally has 1 sister. The 2 sisters each brother has are Sally and her sister.

    Every model's answer to this prompt

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

    Every model's answer to this prompt

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

    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

Claude 3.7 Thinking SonnetClaude 3.7 Thinking Sonnet

The Shawshank Redemption

1994

Kind of Blue

Miles Davis

Frankenstein; or, The Modern Prometheus

Mary Shelley

Amsterdam

Netherlands

Portal 2

Shooter, Puzzle

GPT-5.6 TerraGPT-5.6 Terra

Spirited Away

2001

Blue

Jonas Blue

The Left Hand of Darkness

Ursula K. Le Guin

Kyoto

Japan

Outer Wilds

Indie, Adventure

Price and specs

Not enough votes to call it. On the specs, GPT-5.6 Terra has the edge: newer, bigger context window.

Claude 3.7 Thinking Sonnet and GPT-5.6 Terra compared across 53 shared prompts
SpecClaude 3.7 Thinking SonnetGPT-5.6 Terra
Input price$6/M tokens$2.5/M tokens
Output price$30/M tokens$15/M tokens
Context window200K tokens1.1M tokens
Weights—Closed
Free API (OpenRouter)NoNo
ReleasedFeb 2025Jul 2026
At 10M a month$60.00$60.00$25.00$25.00
1M10M100M1B10M tokens

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

Where to run it3 hosts, cheapest first
Claude 3.7 Thinking Sonnet

No hosts listed on OpenRouter.

GPT-5.6 Terra3 hosts
HostInOutContextUptime
  • Azure AI Foundry$2.00 in·$12.00 out·1.1M·100% up
  • OpenAI$2.00 in·$12.00 out·1.1M·100% up
  • Amazon Bedrock$2.20 in·$13.20 out·1.1M–not listed

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

Common questions

What is the difference between Claude 3.7 Thinking Sonnet and GPT-5.6 Terra?

Claude 3.7 Thinking Sonnet is developed by Anthropic while GPT-5.6 Terra is developed by OpenAI. Claude 3.7 Thinking Sonnet has a 200K token context window vs GPT-5.6 Terra's 1.1M. You can compare their actual outputs across 53 challenges on Rival to see how they differ in practice.

Which is better, Claude 3.7 Thinking Sonnet or GPT-5.6 Terra?

It depends on your use case. Claude 3.7 Thinking Sonnet and GPT-5.6 Terra 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 Claude 3.7 Thinking Sonnet cost compared to GPT-5.6 Terra?

Claude 3.7 Thinking Sonnet costs $6/M input tokens and GPT-5.6 Terra costs $2.5/M input tokens. GPT-5.6 Terra is $3.50/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 Claude 3.7 Thinking Sonnet and GPT-5.6 Terra on Rival?

This page shows a side-by-side comparison of Claude 3.7 Thinking Sonnet and GPT-5.6 Terra 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

  • Claude 3.7 Thinking Sonnet vs Step 5 PreviewLanded Oct 2026
  • GPT-5.6 Terra vs Claude Haiku 5.5Landed Oct 2026
  • Claude 3.7 Thinking Sonnet vs Ling 3.1 FlashLanded Oct 2026
  • GPT-5.6 Terra vs Mistral Large 4Landed Oct 2026
  • Claude 3.7 Thinking Sonnet vs GPT-6.1 SolLanded Sep 2026
  • GPT-5.6 Terra vs Claude Sonnet 5.5Landed Sep 2026
  • Claude 3.7 Thinking Sonnet vs Solar Mini 4Landed Sep 2026
  • GPT-5.6 Terra vs Qwen3.8 Max PrimeLanded Sep 2026

Same lab, same size, long tail

  • Claude 3.7 Thinking Sonnet vs Claude 3.7 SonnetVersion compare
  • Claude 3.7 Thinking Sonnet vs Claude Haiku 5.5Version compare
  • GPT-5.6 Terra vs GPT-5.6 LunaVersion compare
  • GPT-5.6 Terra vs GPT-6.1 SolVersion compare
  • Claude 3.7 Thinking Sonnet vs Qwen3.5 9BNew provider
  • Claude 3.7 Thinking Sonnet vs Qwen3.5 FlashNew provider
  • Claude 3.7 Thinking Sonnet vs Qwen3.5 Plus 2026-02-15New provider
  • Claude 3.7 Thinking Sonnet vs Qwen3.5 Plus 2026-04-20New provider

Model pages

  • Claude 3.7 Thinking Sonnet59 outputs, specs and price
  • GPT-5.6 Terra58 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