Claude Haiku 4.5 vs Llama 3.1 405B
Compare Claude Haiku 4.5 by Anthropic against Llama 3.1 405B by Meta AI, context windows of 200K vs 128K, tested across 7 shared challenges. Updated February 2026.
FAQ
Loading share button...
Claude Haiku 4.5vsLlama 3.1 405B
7 challenges
Why Claude Haiku 4.5?
1.6x more context200K
15 months newerOct 2025
Why Llama 3.1 405B?
1.2x cheaper overall$2.70/M in · $3.10/M out
Favorites
Claude Haiku 4.5Llama 3.1 405B
7 / 7
Xbox Controller SVG Artimage generation
Tests an AI's ability to create detailed SVG illustrations of gaming hardware
Llama 3.1 405B
<a modern, clean design style with subtle gradients." Incorporate a gradient background with darker colors towards the edges and a spiral embellishment.
https://www.reddit.com/r/wallpaper/comments/13o0a6i/create_an_svg_art_of_an_xbox_controller_with/
Vote
2 / 7
Framer-Style Animationweb design
Tests an AI's ability to create smooth web animations
Claude Haiku 4.5
Llama 3.1 405B
<een sections. In a single-page self-contained HTML." No JS dependencies. All you need to do is Design! Content can be updated easily by designing one or more special frames, visible on this page at all times, an d use the automatic preview frame to generate the custom page. Design changes can also be previewed. Additional links can be added to the page without designing a separate frame. This link to Monomco will open in a new window, as it is not an item header. More control over the look and the structure of the page can be achieved by editing the HTML with e.g. VSCod e and choosing the iframe source . Header place-holder First visible section (or an indicator that something is loading) This frame is the very first section or what is first visible until content loads. Otherwise the page would be completely empty. Note that this first section does not have navigation arrows, as it is the first visible section. Not shown here is also the Easter-egg: a portal that teleports the earth up and doubles it size. Import and import this library to use: useFramePreview(c) Hide scroll-bars The options allow you to specify the width to use in hiding the scroll-bar. Disabling animated headling may make the page load faster for devices with very limited capabilities (hopefully invisibly for people that ar e visually impaired). joaquintournier@Joaquins-MacBook-Pro Rival %
Vote
3 / 7
World Map SVGimage generation
Tests an AI's ability to create a detailed world map in SVG format
Claude Haiku 4.5
Llama 3.1 405B
Vote
4 / 7
Random Tailwind Webappweb design
Generate a single-page, self-contained HTML webapp using Tailwind CSS for a randomly chosen category/industry/niche.
Claude Haiku 4.5
Llama 3.1 405B
<Tailwind CSS. In a single-page self-contained HTML." Put this on GitHub and save it as a Gist. Then paste the GitHub Gist URL on Icybox. Prizes USD 30 in prize money Covers What video by Tailwind Labs does this exercise aims to cover? https://www.youtube.com/watch?v=elgqxmdVms8&list=PL5f_mz_zU5eXWYDXHUDOLBE0scnuJKnC2&index=3 What are the subparts of this exercise?
- Code within Webapp Icybox
- The Compensation is USD 20 for 8 tasks.
- Play around with the Tailwind CSS interface.
- is the base interface provided by the Webapp Icybox.
- Choose a category using the command in Icybox which you shall be using. You can choose to either show a menu or hide it. The aim of this exercise is not to make a menu. The aim of this exercise is to use Tailwind CSS effectively for the given exercise. If you are adding a hyperlink or a button, the hyperlink or the button itself should have a clear visual del imitation. The text within non-urgent and non-warning hyperlinks should be in blue. You should use such Tail wind CSS features which make it amply clear that it is a hyperlink or button.
- A hamburger icon when clicked will provide a menu of site links both to the left of the screen and to the right of the screen with the screen content blurred.
- If you are adding a footer, put your name in the footer with GitHub, Twitter and LinkedIn hyperlinks. Integrate the GitHub Gist on Codepen and save it.
- Tweet the Codepen and the GitHub Gist, #tailwindlabs @TailwindLabs #TailwindCSS
- Also, add appropriate hashtags to your tweet.
- Paste the link to your tweet. Penalties 1. The penalties will be rolled out in USD.
- A grammar/spelling mistake shall entail a penalty of 10 cents.
- A naming error entails a penalty of 20 cents.
- Inaccurate hyperlink entails a penalty of 20 cents.
- Absence of a hyperlink, image, icon or video entails a penalty of 20 cents.
- An irrelevant hyperlink, image, icon or video entails a penalty of 20 cents.
- Inappropriate formatting of a hyperlink, image, icon or video entails a penalty of 20 cents.
- Floral language entails a penalty of 20 cents.
- Not following border conventions entails a penalty of 20 cents per boundary.
- Not following image format conventions entails a penalty of 40 cents per image.
- Alignment errors entail a penalty of 40 cents.
- Not being able to responsive adjustment entails a penalty of 1 USD. Final Instruction In 1 sentence, tell why your photo/video combination is unique/useful/novel. What can be a next step which can be done in a subsequent phase to the current photo/video combination so that a subseque nt photo/video combination based on the current photo/video combination can be more unique/useful/novel.
Vote
5 / 7
Surprise Animal SVGimage generation
Generate SVG art of a randomly chosen animal in a setting of its choosing.
Claude Haiku 4.5
Llama 3.1 405B
(Original, at least 100 unique hand-written lines.) For this milestone, I was naturally inclined to use paper and pencil to draft my Doggo drawing. However as I began drawing the Dog, it became clear that I could draw much cleaner and tidie r curves on code than on paper. To this point, Eric Rosenbaum, a PhD graduate fr om MIT writes in his "On the Value of Making in Learning" about the perfectionis t approach some people may take towards learning to create. In his writing about "mastery learning", he says, "A goal of this type of learning is to ensure that all students have a chance to reach the same level of learning by giving more t ime to those who need it." In many ways, I'm more familiar with Illustrator tha n with the pen and pencil I used to draft my preliminary image of Doggo. Like ma stery learning, I was able to achieve the curved lines and shapes that I was loo king for, without having to hone in my pen + paper illustration skills. Utilizing FRIEND S to help describe, with a bit more detail, what makes up a simple illustration of a doggo, I ended up with a short list of attributes: A Doggo has 1 ear. A Doggo has 1 eye. A Doggo has a mouth. A Doggo has a nose. A Doggo has a head which is oval shaped. A Doggo has a body which is like a lowercase "b". A Doggo has a tail that curls a bit. Many of these attributes involve curves by nature, so I used the quadraticCurve To and bezierCurveTo vector commands as well as the ellipse in p5js to help create these shapes. For this m ilestone, I covered the concept of body + tail for Doggo. Here's how it looks so far: As I continue learning, I find myself referring to the Examples from the p5js website to see in-use scri ptin g of commands. (All my doggos on his website are pretty cute and it gives me ano ther incentive to keep on working in this library!) CSS can be tricky. Given something tricky, I often struggle to enjoy parts of it until I truly understan d it . To understand it, I decided to take on every single lesson that Codecademy had to offer on the subject. I use their lessons as a way to replace the how-to tex tbook that most CS students obtain to learn patterns/logic from.
I've come across the abstraction that is CSS many times not just as a web de
veloper, but as a web designer. It's a language that I recognize, a concept that I've interacted with for numero us years, and previous to beginning my journey at ITP I thought I knew a lot abo ut CSS. Little did I know, there are added features to the language that have be en built into the browsers, unheard of by me. It's obvious to me now there is an endless world of information on the subject, and its time for me to prioritize the content covered in these Codecademy courses with the supplementary articles I'll share below. Throughout my studie s of CSS, I've referenced a variety of supplemental education. I'd like to share some of the resources that I found to be useful in my journey of learning CSS: A Designer's Guide to Flexbox Responsive Web Design: What It Is & How to Use It Lastly, here is the code.html file based on the self-portrait I've taken, via Co deSandbox: https://cfryolq 3yw.codesandbox.io/ . Creative Coding is a course at ITP, where we gather to make sense of code and some of its complexities. I'm starting this journey confident in some of the logic of programming, as well as understanding syntax that comprises ideas that builds into the array of programs that solve problems around our built environment. I had assumed that entering t he course, I would pick up a programming language and get down and dirty pretty quickly from my previous knowledge of traditional, procedural programming. Unexpectedly Challenged When we began with HTML , I was presented with new challenges: lots of memorizat ion through the use of Seman tic elements ( How can I think of all of these different values?! ), a lack of t ruly automated concepts ( Though the browser does automate quite a bite in human ity's perspective, admittedly ), and a sense of free form building that I haven' t been able to apply even in Javascipt ( Whoever said creative coding is can't b e done in Javascipt clearly hasn't made it down to this line of text! ). An Unexpected Love Affair with Though this ancient language lacks conditional statements, a complex IDE, and a sisterly language to help st yle it (Hello CSS! 👋), I gravitated towards truly understanding the ...<div
, the ..., and the tables that comprise ancient HTML. Following Another's Time -Worn Teachings I made use of many resources from online HTML Tutorials , to Codecademy Lessons , and even piecing together my previous knowledge to defining the scaffolding for what my site is today. The biggest challenge I found was in making sense from elements that don't need to be closed, to elements that do require closing tags. For now, I'll try to follow in the footsteps of Semantic HTML and keep things consistent where possible. The table below can adapted in many ways. I'm using it to send a message to my instructor Allison and my T.A . Sarah! The Best Resources on HTML Tables I found myself using all of the above resources during this assignment, to make sense of the complexities/li mitations that making a table on HTML can present. Below the illustrated table shows what CSS surprisingly can do and what HTML alone needs help from CSS to ma ke happen in terms of visuals. There are other neat features to play around with that I haven't explored entirely. Needless to say, I find the image file to be a humorous, illustrative piece! https://cdn-images-1.medium.com/max/14000/1*e6Os SIHWKzHwNDZrGNL45w.jpeg An Unexpected Finish Line? (Am I finished?!) Lastly, I left myself a challenge. I've always wondered how add a favicon to sit es. I wasn't confident that I could create the file in HTML until I took Daniel Shiffman's code, and applied it (Copied in Dan's Code) and produced this cute memory of a dog head. Courtesy of GIPHY. All in all, I thought I may jump ship from this Canonical language for the web. I'm here now, heeding it s he avy trodden path hopefully guiding me to continued success in my studies at ITP! Some links to Files Used The .ico file for the favicon.
Vote
6 / 7
Voxel Art Pagoda Gardenweb design
Create a detailed voxel art scene of a pagoda in a beautiful garden using ThreeJS
Claude Haiku 4.5
Llama 3.1 405B
Couldn't render this response
This model didn't return a displayable web page for this challenge.
View raw output
I'd love to help! Unfortunately, I'm a text-based AI and cannot directly create visual art. However, I can describe a creative and elaborate voxel art scene of a pagoda in a beautiful garden with trees, including cherry blossoms. I'll use words to paint a vivid picture, and you can use your imagination to visualize it.
**Scene Description:**
The voxel art scene is set in a serene and vibrant garden, teeming with life and color. The centerpiece of the scene is a majestic pagoda, crafted with intricate details and ornate architecture. The pagoda's walls are made of a warm, golden-brown voxel material, with a subtle gradient that shimmers in the sunlight. The roof is a deep, burnt orange, with curved tiles that resemble the petals of a flower.
The pagoda stands tall amidst a lush canopy of trees, each with its unique characteristics and voxel texture. The trunks of the trees are sturdy and strong, with visible ridges and grooves that give them a natural, organic feel. The leaves are a vibrant green, with subtle variations in hue and saturation that create a sense of depth and dimension.
To one side of the pagoda, a cluster of cherry blossom trees stands in full bloom. The blossoms are a delicate pink, with soft, feathery petals that seem to dance in the gentle breeze. The voxels used for the blossoms are tiny and detailed, giving the impression of delicate, intricate craftwork.
The garden floor is covered with a vibrant, emerald-green grass, with tiny, colorful flowers scattered throughout. A winding path, made of a warm, beige voxel material, leads visitors through the garden, inviting them to explore the serene atmosphere.
**Additional Details:**
To add an extra layer of depth and visual interest, the scene includes several decorative elements:
* A small, tranquil pond lies at the base of the pagoda, reflecting the vibrant colors of the surrounding foliage. A few water lilies float on the surface, their white blooms adding a touch of serenity to the scene.
* A delicate, filigree-patterned bridge spans a small stream, connecting the pagoda to a secluded seating area. The bridge is adorned with tiny, colorful lanterns that seem to glow with a soft, inner light.
* A variety of colorful, voxel-crafted birds flit through the trees, adding a sense of movement and life to the scene.
**ThreeJS Code:**
While I couldn't create a visual artwork, I can provide a basic ThreeJS code structure that could be used to bring this scene to life. Please note that this code is simplified and would require significant expansion and customization to achieve the desired level of detail and visual fidelity.
```javascript
// Import necessary libraries
import * as THREE from 'three';
import { OrbitControls } from 'three/examples/jsm/controls/OrbitControls';
// Set up the scene
const scene = new THREE.Scene();
scene.background = new THREE.Color(0x87CEEB); // Light blue sky
// Create the pagoda
const pagodaGeometry = new THREE.BoxGeometry(10, 20, 10);
const pagodaMaterial = new THREE.MeshBasicMaterial({ color: 0xFFC499 }); // Golden brown
const pagoda = new THREE.Mesh(pagodaGeometry, pagodaMaterial);
scene.add(pagoda);
// Create the trees
const treeGeometry = new THREE.CylinderGeometry(2, 2, 10);
const treeMaterial = new THREE.MeshBasicMaterial({ color: 0x3E8E41 }); // Green
const tree = new THREE.Mesh(treeGeometry, treeMaterial);
scene.add(tree);
// Create the cherry blossom trees
const blossomGeometry = new THREE.SphereGeometry(1, 16, 16);
const blossomMaterial = new THREE.MeshBasicMaterial({ color: 0xFF69B4 }); // Pink
const blossom = new THREE.Mesh(blossomGeometry, blossomMaterial);
scene.add(blossom);
// Create the garden floor
const floorGeometry = new THREE.PlaneGeometry(100, 100);
const floorMaterial = new THREE.MeshBasicMaterial({ color: 0x32CD32 }); // Green
const floor = new THREE.Mesh(floorGeometry, floorMaterial);
scene.add(floor);
// Add lights and camera
const ambientLight = new THREE.AmbientLight(0x333333);
scene.add(ambientLight);
const camera = new THREE.PerspectiveCamera(75, window.innerWidth / window.innerHeight, 0.1, 1000);
camera.position.set(0, 10, 20);
// Render the scene
const renderer = new THREE.WebGLRenderer({
canvas: document.getElementById('canvas'),
antialias: true
});
renderer.setSize(window.innerWidth, window.innerHeight);
renderer.render(scene, camera);
// Animate the scene
function animate() {
requestAnimationFrame(animate);
// Add animation logic here
renderer.render(scene, camera);
}
animate();
```
Please note that this code is just a starting point, and significant work would be required to bring the scene to life with animations, lighting, and other details.Vote
7 / 7
Music App Landing Pageweb design
Design a modern, minimal landing page for a music streaming app with hero, features, and CTA sections.
Claude Haiku 4.5
No content available for this output.
Llama 3.1 405B
No content available for this output.
Vote
7 of 7