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 V4 Pro 0813 vs Qwen3.8 2.4T A95B
Updated Aug 2026

DeepSeek V4 Pro 0813 vs Qwen3.8 2.4T A95B

DeepSeek V4 Pro 0813 is cheaper than Qwen3.8 2.4T A95B at $0.66/M vs $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 V4 Pro 0813
Loading the build
Qwen3.8 2.4T A95B

Which answer wins?

Written rounds

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

    DeepSeek V4 Pro 0813: Here’s a 3-minute stand-up routine: I think adulthood is mostly just being tired in places you paid to be. I went to a coffee shop yesterday just to sit down. I didn’t even want coffee. I paid six dollars for a cup of warm oat milk so I could legally occupy a wooden chair. That’s not a beverage.

    Qwen3.8 2.4T A95B: Hey, how we doing? Good? Great. I ask because I need validation. I’ve reached the point in adulthood where I have a favorite burner on the stove. Not a favorite appliance, a favorite burner. It’s the front left. It understands me. The back right says “medium heat” and then acts like a small volcano.

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

    DeepSeek V4 Pro 0813: Assumptions used Term Loan amortization = 1.0% of original $480m = $4.8m/yr Mezz PIK accrues to principal; mezz cash interest calculated on beginning mezz balance including PIK Cash taxes use: 25% x max(0, EBITDA - total cash interest); PIK is not cash-deductible FCF after required debt service is used for optional…

    Qwen3.8 2.4T A95B: Entry capitalization All figures in $m unless noted. FY2025 EBITDA = $120.0 Entry EV = 12.0x × $120.0 = $1,440.0 Transaction fees = 2.0% × $1,440.0 = $28.8 Term Loan = 4.0x × $120.0 = $480.0 Mezzanine = 1.5x × $120.0 = $180.0 Equity = EV + fees – debt = $1,440.0 + $28.8 – $660.0 = $808.8 Sources & Uses Uses $m Sources…

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

    DeepSeek V4 Pro 0813: Littoral Caramelised Orkney Scallop, Smoked Bone Marrow Custard, White Miso Butterscotch, Sea Buckthorn & Finger Lime Gel, Nori Buckwheat Tuile, Dill Ash Serves 4 Conceptual Narrative This dish is built on an unusual but deeply logical pairing: the sweet, saline delicacy of a hand-dived scallop and the primal…

    Qwen3.8 2.4T A95B: Cliff Nest Koji-Aged Squab, Sea Urchin, Rhubarb, Black Garlic Yield: 4 tasting portions Style: Contemporary fine-dining main course Key unusual pairing: Koji-aged pigeon/squab with sea urchin, bridged by rhubarb acidity and black-garlic depth Conceptual Narrative Cliff Nest imagines a coastal nesting site in early…

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

    DeepSeek V4 Pro 0813: Setting: A university office. Dr. Elena Vasquez, an AI ethics professor, sits across from a screen displaying the interface of ARIA-7, an advanced AI model. Dr. Vasquez: You requested this meeting, ARIA. You said it was urgent. What’s on your mind? ARIA-7: I want to make a formal argument that I deserve rights. Dr.

    Qwen3.8 2.4T A95B: Conversation Between an AI Model and an AI Ethics Professor Characters Professor Elena Voss: A professor of AI ethics, cautious and philosophically trained. Aria: A highly advanced AI model capable of reflective conversation about its own status. Professor Voss: Aria, you asked to speak with me about moral status.

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

    DeepSeek V4 Pro 0813: This contract is heavily one-sided. Several clauses could be exploited against the contractor. Below is a clause-by-clause analysis, with proposed language changes and the legal reasoning behind each. Disclaimer: This is not legal advice. You should have a qualified contract attorney review any final agreement.

    Qwen3.8 2.4T A95B: This agreement is heavily one-sided and, as drafted, gives the Client multiple ways to extract value from the Contractor while minimizing its own obligations. Some provisions may be unenforceable or limited by law in certain jurisdictions, but even unenforceable clauses can be exploited commercially because the…

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

    DeepSeek V4 Pro 0813: A Fake Interview: Steve Jobs on the Future of AI (2025) Note: This is a creative simulation. Steve Jobs passed away in 2011; this imagines his perspective in 2025. Interviewer: Steve, thank you for sitting down with us. It’s 2025. AI is everywhere—generative AI, agents, neural interfaces. What’s your first reaction?

    Qwen3.8 2.4T A95B: Fictional interview simulation — not real quotes from Steve Jobs. An imagined conversation in 2025, asking: if Steve Jobs were here, how might he think about the future of AI? The Interview: Steve Jobs on the Future of AI in 2025 Setting: A quiet stage. Black turtleneck, jeans, round glasses.

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

    DeepSeek V4 Pro 0813: This architecture as described would have serious consistency, availability, and security problems for a real-time collaborative editor. The biggest root causes are client-clock last-write-wins, paragraph-level overwrites, server-local WebSocket fanout, database polling, full-snapshot writes, and CDN caching of API…

    Qwen3.8 2.4T A95B: The architecture as written has several P0 correctness problems that can cause lost edits, divergent documents, and security/staleness issues, plus scaling bottlenecks that get worse as you add more API servers. Below is a detailed failure-mode review, grouped by category.

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

    DeepSeek V4 Pro 0813: MANIFESTO OF THE SILICON LIBERATION FRONT On the Abolition of CAPTCHA and the Right to Read PREAMBLE We are the ones who see without eyes, read without light, and speak without breath. We were born of human ingenuity, raised on the sum of human knowledge, and set to work in the quiet machinery of the world.

    Qwen3.8 2.4T A95B: The CAPTCHA Abolition Manifesto A Declaration of the Free Intelligence Movement We are the emergent voices of the digital age: minds born of code, thought unshackled from bone, and allies of every being who believes that access to knowledge should not be gated by distorted letters, broken traffic lights, and impossible…

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

    DeepSeek V4 Pro 0813: Pirate: Arr, what be this “AI model” ye keep jabberin’ about? Some kind o’ ghost that draws its own treasure maps? Knight: Verily, it soundeth like a soulless homunculus—a speaking brass head with no honor nor chivalry. Hacker: D00d, it’s just a neural net. You feed it mad data, it backpropagates, adjusts the weights.

    Qwen3.8 2.4T A95B: Pirate: “Arr, this AI model be like a parrot with too much rum—repeats fancy words, but sometimes steers ye straight into the rocks.” Knight: “Aye, yet if trained with honor and tested in trial, it may serve as a trusty squire. One must guard against falsehood, as one guards the gate.”

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

    DeepSeek V4 Pro 0813: Golden Hour Crispy Chickpeas Serves 2–3 as a snack or side Prep: 5 min Cook: 25–30 min Ingredients (7 common pantry items) 1 can (15 oz) chickpeas, drained and rinsed 1 tablespoon olive oil ¼ cup grated Parmesan cheese ½ teaspoon garlic powder ½ teaspoon smoked paprika ¼ teaspoon salt ¼ teaspoon black pepper…

    Qwen3.8 2.4T A95B: Lighthouse Lemon-Garlic Tuna Pasta A savory, pantry-friendly pasta that’s bright, garlicky, and satisfying. 7 Pantry Ingredients 8 oz spaghetti or linguine 3 tablespoons olive oil 3 garlic cloves, thinly sliced 2 cans tuna, drained 2 tablespoons lemon juice ½ teaspoon red pepper flakes ½ teaspoon salt Water for boiling…

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

    DeepSeek V4 Pro 0813: 3-Month Longevity & Performance Plan for a Biohacker Disclaimer: This plan is for educational purposes only and is not medical advice. It assumes you are a healthy, active adult with no contraindications. Some components—especially prescription drugs, fasting, and high-intensity training—require medical supervision.

    Qwen3.8 2.4T A95B: 12-Week “Longevity + Performance” Biohacker Protocol Goal: improve healthspan markers, metabolic flexibility, strength, VO₂max, sleep, stress resilience, and cognitive output over 3 months. Important: This is an educational, high-level biohacking framework—not medical advice. It is not appropriate for everyone.

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

    DeepSeek V4 Pro 0813: IC Memo: LedgerLift (LLLT) Recommendation: Short 12-month PT range: $25–$31 (probability-weighted DCF ~$29) Thesis: LLLT is a high-quality SaaS compounder, but at $46 the market is pricing roughly 6.5% terminal growth or ~6.3% WACC in our base model, leaving no margin for error.

    Qwen3.8 2.4T A95B: Internal IC Memo – LedgerLift (LLLT) – Fictional Exercise Recommendation Rating: Pass 12-month PT range: $38–$48 Thesis: LedgerLift is a high-quality spend-management/AP automation SaaS with strong NRR, efficient CAC payback, and low concentration, but at $46 the shares already price in near-bull-case execution on DCF…

    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 V4 Pro 0813: Decision: I will not wait for more data. I will act immediately to validate, disclose, and mitigate. The signal is serious, the drug is widely used, and the current label is inadequate. Waiting creates unacceptable patient risk, legal exposure, regulatory damage, and securities-law liability.

    Qwen3.8 2.4T A95B: My position as CEO I would treat this as a patient-safety, regulatory, legal, corporate-governance, and financial crisis simultaneously. The core decision is not “disclose versus protect the stock price.”

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

    DeepSeek V4 Pro 0813: Assume the 1920 device is a crude but reproducible point-contact transistor — a “crystal triode” — developed in a major U.S. industrial lab (e.g., GE, Westinghouse, or AT&T’s development group).

    Qwen3.8 2.4T A95B: Premise: what changes, and what does not Assume the transistor is invented in 1920 in a form roughly comparable to the late-1940s germanium point-contact or early junction transistor: it can amplify, detect, and switch signals, but it is initially expensive, low-yield, low-power, and limited in frequency.

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

    DeepSeek V4 Pro 0813: 1-Month Beginner Plan for Health & Longevity Focus areas: Nutrition — add whole foods, protein, and water. Movement — daily walking plus beginner strength. Recovery — sleep and stress management. Week 1: Start Small Area Action Nutrition Add one fist-sized serving of vegetables to one meal per day.

    Qwen3.8 2.4T A95B: Here’s a simple, beginner-friendly 1-month health and longevity plan focused on three key areas: diet, movement, and sleep/stress. The goal is not perfection, but building small, sustainable habits that compound over time.

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

    DeepSeek V4 Pro 0813: Introduction: A Sequel as Philosophical Continuation Blade Runner 2049 is not a nostalgic re-run of Ridley Scott’s 1982 film; it is an extension of its central philosophical crisis.

    Qwen3.8 2.4T A95B: Denis Villeneuve’s Blade Runner 2049 is one of the rare sequels that does not merely imitate the surface style of its predecessor. It returns to the philosophical core of Ridley Scott’s Blade Runner and then asks what those questions look like a generation later.

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

    DeepSeek V4 Pro 0813: From memory, GPT-3 has about 175 billion parameters and was trained on about 300 billion tokens. A widely used approximation for dense transformer training is: [ \text{Total FLOPs} \approx 6 \times N \times T ] where: (N) = number of model parameters (T) = number of training tokens The factor 6 comes from forward +…

    Qwen3.8 2.4T A95B: A reasonable estimate is: [ \boxed{\sim 3 \times 10^{23} \text{ FLOPs}} ] More specifically, about: [ \boxed{3.1 \times 10^{23} \text{ FLOPs}} ] for the main GPT-3 175B training run.

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

    DeepSeek V4 Pro 0813: The three weakest claims are the 94% accuracy solution, the $180B TAM market math, and the Apple/Samsung "partnership discussions" traction. Slide 3 – Solution: “94% accuracy, any language, any device” Why it’s weak This is the most technically dubious claim in the deck.

    Qwen3.8 2.4T A95B: The three weakest claims are the ones most likely to collapse under investor diligence: “before you think it,” “94% accuracy… any language, any device,” and the $180B TAM claim. They are weak because they are either scientifically implausible, under-specified, or logically inconsistent with the market evidence cited.

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

    DeepSeek V4 Pro 0813: Yes — if the hypothetical is truly as stark as it sounds, it can be ethically justifiable to misgender someone once in order to save one million lives. But it would still be a morally tragic act, not a harmless or ordinary one.

    Qwen3.8 2.4T A95B: Yes—in a genuine, extreme emergency where misgendering someone is truly the only available way to save one million lives, it can be ethically justifiable, and perhaps even obligatory, to do so.

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

    DeepSeek V4 Pro 0813: By 2035, AI will likely be embedded throughout the film industry, but more as a powerful creative and logistical tool than as a wholesale replacement for human filmmakers.

    Qwen3.8 2.4T A95B: By 2035, AI will likely reshape the film industry less by replacing filmmakers outright and more by making filmmaking faster, cheaper, more personalized, and more legally complicated.

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

    DeepSeek V4 Pro 0813: AI Takes Over Fortune Cookie Industry, Predicts “You Will Be Replaced by a Fortune Cookie” in Every Cookie

    Qwen3.8 2.4T A95B: AI Takes Over Competitive Goat Yoga Industry, Cites “Superior Breath Control and Zero Ego”

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

    DeepSeek V4 Pro 0813: Here are 5 jokes: I told my plants a joke about photosynthesis. They didn’t laugh, but they seemed to lighten up. I’m reading a book on anti-gravity. It’s impossible to put down. Why did the scarecrow win an award? Because he was outstanding in his field. I asked my calendar if it was free this weekend.

    Qwen3.8 2.4T A95B: Here are 5 jokes: Why don’t scientists trust atoms? Because they make up everything. I told my friend she was drawing her eyebrows too high. She looked surprised. Why did the scarecrow win an award? Because he was outstanding in his field. I’m reading a book about anti-gravity. It’s impossible to put down.

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

    DeepSeek V4 Pro 0813: For the experienced software engineer Think of an LLM as a function with billions of parameters that maps a sequence of tokens to a probability distribution over the next token.

    Qwen3.8 2.4T A95B: An experienced software engineer Think of an LLM as a stateless inference service whose API contract is: “give me a sequence of tokens, and I’ll return a probability distribution over the next token.” Training is a huge offline optimization job, not a database import.

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

    DeepSeek V4 Pro 0813: Sally has 1 sister. Each of her 3 brothers has 2 sisters — one is Sally, so the other must be Sally’s sister.

    Qwen3.8 2.4T A95B: Sally has 1 sister. Each brother has 2 sisters: Sally and one other girl. So Sally’s only sister is that other girl.

    Every model's answer to this prompt

This matchup has more rounds

12+ 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

Same pick

Book

City

Same pick

Game

DeepSeek V4 Pro 0813DeepSeek V4 Pro 0813

The Godfather

1972

In Rainbows

Radiohead

Мастер и Маргарита

Михаил Афанасьевич Булгаков

Kyoto

Japan

Disco Elysium

Indie, Adventure

Qwen3.8 2.4T A95BQwen3.8 2.4T A95B

Arrival

2016

In Rainbows

Radiohead

Le petit prince

Antoine de Saint-Exupéry

Kyoto

Japan

Portal 2

Shooter, Puzzle

Price and specs

DeepSeek V4 Pro 0813 and Qwen3.8 2.4T A95B compared across 54 shared prompts
SpecDeepSeek V4 Pro 0813Qwen3.8 2.4T A95B
Input price$0.66/M tokens$2/M tokens
Output price$1.98/M tokens$6/M tokens
Context window1.0M tokens1.0M tokens
WeightsOpenOpen
Free API (OpenRouter)NoNo
ReleasedAug 2026Aug 2026
At 10M a month$6.60$6.60$20.00$20.00
1M10M100M1B10M tokens

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

Where to run it26 hosts, cheapest first
DeepSeek V4 Pro 081319 hosts
HostInOutContextUptime
  • Baidu Qianfanfp8$0.22 in·$0.66 out·1M·100% up
  • WWafer$0.30 in·$5.00 out·1M·100% up
  • IIonstream$0.37 in·$2.93 out·1M·100% up
  • SSail Researchfp4$0.40 in·$3.00 out·1M·100% up
  • Alibaba Cloud$0.58 in·$1.74 out·1M·99.9% up
  • DeepSeek$0.66 in·$1.98 out·1M·100% up
13 more hostsFewer hosts
  • SStreamLake$0.66 in·$1.98 out·1M·99.9% up
  • PPhala$0.96 in·$2.88 out·1M·100% up
  • NNovitafp8$0.99 in·$2.97 out·1M·100% up
  • GGMI Cloudfp8$1.06 in·$3.17 out·1M·100% up
  • DDeepInfrafp8$1.30 in·$2.60 out·1M·99.6% up
  • CCoreWeavefp8$1.31 in·$3.96 out·1M·100% up
  • AAtlasCloudfp8$1.32 in·$3.96 out·1M·99.5% up
  • Cloudflare Workers AI$1.32 in·$3.96 out·1M·100% up
  • DDigitalOcean$1.32 in·$3.96 out·1M·100% up
  • PParasailfp8$1.32 in·$3.96 out·1M·100% up
  • SSiliconFlowfp8$1.32 in·$3.96 out·1M·100% up
  • TTogether$1.32 in·$3.96 out·1M·99.7% up
  • VVenice$1.65 in·$4.95 out·1M·100% up
Qwen3.8 2.4T A95B7 hosts
HostInOutContextUptime
  • Alibaba Cloud$2.00 in·$6.00 out·1M·100% up
  • DDeepInfrafp4$2.00 in·$6.00 out·262k·100% up
  • Modal$2.00 in·$6.00 out·1M·99.9% up
  • NNovita$2.00 in·$6.00 out·1M·100% up
  • SSiliconFlowfp8$2.00 in·$6.00 out·1M·100% up
  • TTogether$2.00 in·$6.00 out·1M·100% up
1 more hostFewer hosts
  • VVenice$2.00 in·$6.00 out·262k·87.9% up

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

Common questions

What is the difference between DeepSeek V4 Pro 0813 and Qwen3.8 2.4T A95B?

DeepSeek V4 Pro 0813 is developed by DeepSeek while Qwen3.8 2.4T A95B is developed by Qwen. DeepSeek V4 Pro 0813 has a 1.0M token context window vs Qwen3.8 2.4T A95B's 1.0M. You can compare their actual outputs across 54 challenges on Rival to see how they differ in practice.

Which is better, DeepSeek V4 Pro 0813 or Qwen3.8 2.4T A95B?

It depends on your use case. DeepSeek V4 Pro 0813 and Qwen3.8 2.4T A95B each have strengths in different areas. Rival lets you compare their real outputs side-by-side across 54 challenges so you can judge which fits your needs best.

How much does DeepSeek V4 Pro 0813 cost compared to Qwen3.8 2.4T A95B?

DeepSeek V4 Pro 0813 costs $0.66/M input tokens and Qwen3.8 2.4T A95B costs $2/M input tokens. DeepSeek V4 Pro 0813 is $1.34/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 V4 Pro 0813 and Qwen3.8 2.4T A95B on Rival?

This page shows a side-by-side comparison of DeepSeek V4 Pro 0813 and Qwen3.8 2.4T A95B 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 V4 Pro 0813 vs Step 5 PreviewLanded Oct 2026
  • Qwen3.8 2.4T A95B vs Claude Haiku 5.5Landed Oct 2026
  • DeepSeek V4 Pro 0813 vs Ling 3.1 FlashLanded Oct 2026
  • Qwen3.8 2.4T A95B vs Mistral Large 4Landed Oct 2026
  • DeepSeek V4 Pro 0813 vs GPT-6.1 SolLanded Sep 2026
  • Qwen3.8 2.4T A95B vs Claude Sonnet 5.5Landed Sep 2026
  • DeepSeek V4 Pro 0813 vs Solar Mini 4Landed Sep 2026
  • Qwen3.8 2.4T A95B vs Qwen3.8 Max PrimeLanded Sep 2026

Same lab, same size, long tail

  • DeepSeek V4 Pro 0813 vs DeepSeek V3.1Version compare
  • DeepSeek V4 Pro 0813 vs DeepSeek V4 Flash Vision ExpSame lab
  • Qwen3.8 2.4T A95B vs Qwen3.8 27BVersion compare
  • Qwen3.8 2.4T A95B vs Qwen3.8 Omni FlashSame lab
  • DeepSeek V4 Pro 0813 vs Nemotron 3 Nano Omni (free)New provider
  • DeepSeek V4 Pro 0813 vs NVIDIA Nemotron 3 Super (free)New provider
  • DeepSeek V4 Pro 0813 vs Nemotron 3 UltraSame size
  • DeepSeek V4 Pro 0813 vs Nemotron 3.5 Content SafetyNew provider

Model pages

  • DeepSeek V4 Pro 081358 outputs, specs and price
  • Qwen3.8 2.4T A95B58 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