01svg-set
simonw/pelican-bicycle
The original repository and early SVG outputs.
Interface em português. Títulos, descrições, prompts e documentos dos autores são preservados no idioma da fonte ou na versão em inglês do arquivo.
PROVENANCE
Sources & repositories
Repositories & demos
23 entry pointsOriginal licenses apply; check each repository’s LICENSE. Unverified source packages are not available for download.
02threejs-game
Licoy/tihuqiche
An island cycling game built with Vue 3 and Three.js.
03svg-set
scosman/pelicans_riding_bicycles
A curated gallery of model-generated SVGs.
04eval-harness
haowang02/auto-pelican
An agent that generates and evaluates SVG animations.
05blender
simonw/gpt-6-astra-blender-pelican-bicycle
A cycling pelican built in local Blender with Astra.
06app
simonw/pedalican
A cycling pelican desktop pet for Codex.
07svg-grid
JoJohanse/pelican-bicycle-benchmark
SVG outputs compared across multiple models.
08agent
youngbrioche/agentic-pelican-on-a-bicycle
Iterative pelican drawing with an agent.
09html-anim
xiiiblue/pelican-bicycle
HTML animations compared across models.
10svg-animated
taikongren776/pelican-bicycle
A seaside cycling scene in animated SVG.
11set
XinhanWang/bicycle-pelican
A collection exploring bicycles and pelicans.
12svg-animated
Superxian-boop/pelican-bicycle
Cycling expressed in pure SVG animation.
13webapp
heysong20651901/pelican-bike
An SVG-based pelican web app.
14svg-set
sebseb7/pelicanTest
A collection of model-generated SVG experiments.
15threejs-game
claude-opus-5-5.riba2534.cn
A playable coastal 3D scene with waves and cycling.
16threejs-game
tihuqiche.com
The online edition of the open-source cycling game.
174kb-demo
demobench.janekm.com
A 4 KB demo with GIF, assembly source and a WASM runtime.
18html-anim
CHNAICORP/pelican-ride
A single-file SVG and HTML animation.
19html-set
vastxie/pelican-gallery-data
Original HTML animations and model runs; 164 extracted files.
20svg-animated
Vulcan575/pelican-riding-bicycle
A dependency-free teal city bicycle animation.
21html-grid
RustFisher · 3-model comparison
Astra, Gemini 3.8 Flash and DeepSeek 4.1 Flash compared.
22threejs
ohmyopus · 3D city
An archived preview with Python / pybevy source.
23html-anim
prab · upstream unverified
Attributed to DeepSeek V4.1 Flash; upstream repository unverified.
Why does this pelican lead back to Simon?
On October 25, 2024, Simon published SVG outputs from 16 models using the same prompt. He chose pelicans because he liked them and believed bicycle-riding pelican SVGs were then rare in training data.
Generate an SVG of a pelican riding a bicycleThe community extended the prompt into animation, video, 3D and games; original sources remain indexed. The museum focuses on code-generated works across presentation media; direct text-to-video is outside the main collection. The timeline follows individual works and source-reported model labels. Different generation conditions and single samples do not establish overall capability.
Read the original experiment ↗Simon’s related writing
144 entries- Claude Opus 5.5, GPT-6 Sol, GPT-6 Luna, and a new price war
- The Creative Spirit of Who Framed Roger Rabbit
- Introducing GPT-6 Astra for developers
- TIL Using Blender with coding agents on macOS I've been having fun with Blender in ChatGPT Codex on my Mac recently. Get
- The Pelican comparison grid for Astra is pretty interesting
- Release llm-gemini 0.34 New model gemini-3.8-flash for Gemini 3.8 Flash , with low, medium and high thinking levels. #14
- Claude Fable 5.1 made me a really nice animated pelican
- Introducing Hy4 Preview
- Qwen3.8-Flash-Next
- Qwen 3.8 27B is excellent, but it defaults to wildly overthinking things
- Release llm-gemini 0.33 It's been a while since the last llm-gemini release. This version of the plugin adds support for
- DeepSeek V4 Pro 0813 (on OpenRouter)
- Introducing Muse Glimmer
- Introducing Muse Code and Muse Spark 1.2
- deepseek-ai/DeepSeek-V4-Flash-0731
- Are AI labs pelicanmaxxing?
- Who’s Afraid of Chinese Models?
- Kimi K3, and what we can still learn from the pelican benchmark
- Inkling: Our open-weights model
- simonw/pedalican
- The new GPT-5.6 family: Luna, Terra, Sol
- Introducing Muse Spark 1.1
- tencent/Hy3
- What’s new in Claude Sonnet 5
- Ornith-1.0: Self-Scaffolding LLMs for Agentic Coding
- GLM-5.2 is probably the most powerful text-only open weights LLM
- DiffusionGemma
- Initial impressions of Claude Fable 5
- Claude Opus 4.8: “a modest but tangible improvement”
- Gemini 3.5 Flash: more expensive, but Google plan to use it for everything
- The last six months in LLMs in five minutes
- Granite 4.1 3B SVG Pelican Gallery
- @scottjla on Twitter in reply to my pelican riding a bicycle benchmark: I feel like we need to stack these tests now I c
- DeepSeek V4—almost on the frontier, a fraction of the price
- A pelican for GPT-5.5 via the semi-official Codex backdoor API
- Qwen3.6-27B: Flagship-Level Coding in a 27B Dense Model
- scosman/pelicans_riding_bicycles
- Release llm-openrouter 0.6 llm openrouter refresh command for refreshing the list of available models without waiting fo
- Qwen3.6-35B-A3B on my laptop drew me a better pelican than Claude Opus 4.7
- Meta’s new model is Muse Spark, and meta.ai chat has some interesting tools
- GLM-5.1: Towards Long-Horizon Tasks
- Gemma 4: Byte for byte, the most capable open models
- GPT-5.4 mini and GPT-5.4 nano, which can describe 76,000 photos for $52
- Introducing Mistral Small 4
- Introducing GPT‑5.4
- Gemini 3.1 Flash-Lite
- Gemini 3.1 Pro
- Introducing Claude Sonnet 4.6
- Opus rear bike frame makes no sense structurally, there is no connection from pedals to rear wheel, both legs of Pelican
- Qwen3.5: Towards Native Multimodal Agents
- Introducing GPT‑5.3‑Codex‑Spark
- Gemini 3 Deep Think
- GLM-5: From Vibe Coding to Agentic Engineering
- Two major new model releases today, within about 15 minutes of each other. Anthropic released Opus 4.6 . Here's its peli
- Kimi K2.5: Visual Agentic Intelligence
- 2025: The year in LLMs
- Introducing GPT-5.2-Codex
- Gemini 3 Flash
- GPT-5.2
- Devstral 2
- DeepSeek-V3.2
- LLM SVG Generation Benchmark
- Claude Opus 4.5, and why evaluating new LLMs is increasingly difficult
- Olmo 3 is a fully open LLM
- Building more with GPT-5.1-Codex-Max
- Trying out Gemini 3 Pro with audio transcription and a new pelican benchmark
- Introducing GPT-5.1 for developers
- What happens if AI labs train for pelicans riding bicycles?
- Agentic Pelican on a Bicycle
- Pelican on a Bike—Raytracer Edition
- Reverse engineering Codex CLI to get GPT-5-Codex-Mini to draw me a pelican
- Kimi K2 Thinking
- Introducing SWE-1.5: Our Fast Agent Model
- MiniMax M2 & Agent: Ingenious in Simplicity
- Composer: Building a fast frontier model with RL
- Visual Features Across Modalities: SVG and ASCII Art Reveal Cross-Modal Understanding
- Introducing Claude Haiku 4.5
- gpt-image-1-mini
- GPT-5 pro
- Two new models from Chinese AI labs in the past few days. I tried them both out using llm-openrouter : DeepSeek-V3.2-Exp
- Claude Sonnet 4.5 is probably the “best coding model in the world” (at least for now)
- Improved Gemini 2.5 Flash and Flash-Lite
- GPT-5-Codex
- Grok 4 Fast
- GPT‑5-Codex and upgrades to Codex
- Qwen3-Next-80B-A3B
- Kimi-K2-Instruct-0905
- DeepSeek 3.1
- David Ho on BlueSky: A pelican tried to eat my bike
- Introducing Gemma 3 270M: The compact model for hyper-efficient AI
- Qwen3-4B-Thinking: “This is art—pelicans don’t ride bikes!”
- A couple of weeks ago I was invited to OpenAI's headquarters for a "preview event", for which I had to sign both an NDA
- GPT-5: Key characteristics, pricing and model card
- OpenAI’s new open weight (Apache 2) models are really good
- Claude Opus 4.1
- XBai o4
- Deep Think in the Gemini app
- Trying out Qwen3 Coder Flash using LM Studio and Open WebUI and LLM
- Qwen3-30B-A3B-Thinking-2507
- My 2.5 year old laptop can write Space Invaders in JavaScript now, using GLM-4.5 Air and MLX
- GLM-4.5: Reasoning, Coding, and Agentic Abililties
- Qwen3-235B-A22B-Thinking-2507
- Instagram Reel: Veo 3 paid preview
- Qwen3-Coder: Agentic Coding in the World
- Qwen/Qwen3-235B-A22B-Instruct-2507
- moonshotai/Kimi-K2-Instruct
- Grok 4
- Infinite Monkey
- Introducing Gemma 3n: The developer guide
- Mistral-Small 3.2
- Trying out the new Gemini 2.5 model family
- o3-pro
- Magistral — the first reasoning model by Mistral AI
- The last six months in LLMs, illustrated by pelicans on bicycles
- Live blog: Claude 4 launch at Code with Claude
- Gemini 2.5: Our most intelligent models are getting even better
- Tucked into today's Google I/O keynote, a blink-and-you'll miss it moment : The pelican in the keynote was created by Al
- In addition to my workshop the other day I'm also participating in the poster session at PyCon US this year. This means
- Gemini 2.5 Pro Preview: even better coding performance
- Qwen 3 offers a case study in how to effectively release a model
- Start building with Gemini 2.5 Flash
- GPT-4.1: Three new million token input models from OpenAI, including their cheapest model yet
- Putting Gemini 2.5 Pro through its paces
- deepseek-ai/DeepSeek-V3-0324
- OpenAI platform: o1-pro
- mlx-community/OLMo-2-0325-32B-Instruct-4bit
- Notes on Google’s Gemma 3
- Initial impressions of GPT-4.5
- Claude 3.7 Sonnet and Claude Code
- Andrej Karpathy’s initial impressions of Grok 3
- Gemini 2.0 is now available to everyone
- DeepSeek-R1 and exploring DeepSeek-R1-Distill-Llama-8B
- microsoft/phi-4
- Gemini 2.0 Flash “Thinking mode”
- Veo 2
- <model-viewer> Web Component by Google
- Sora
- I can now run a GPT-4 class model on my laptop
- New Gemini model: gemini-exp-1206
- QwQ: Reflect Deeply on the Boundaries of the Unknown
- Say hello to gemini-exp-1121
- Recraft V3
- Qwen2.5-Coder-32B is an LLM that can code well that runs on my Mac
- Pelicans on a bicycle
No matching articles.
Scroll within the list to browse. These reading references are not counted as artworks.
Reading notes & counting
Follow each work back to its code, creator and original question. This reading index of sources and repositories is not an artwork count. Animation, video, 3D, games and agent projects retain generation conditions, iterations and human involvement.
1165 independent works; the 1012 classic works on the timeline are a subset, not an additional total. The focus is LLM code-generated work across all dates and presentation media. Direct text-to-video is excluded; recordings of code-generated animation remain eligible. Model labels and attribution are source-reported, not independently authenticated.