AI Resources
AI resources
The routers, weights, leaderboards, docs and trackers the desk itself works from. Each card opens the resource in a new tab.
23 resources · 7 groups · every link checked 15 October 2026
Use the models
OpenRouter One API in front of every major model, with live pricing, latency and usage rankings. The fastest way to compare frontier and open-weight models side by side.openrouter.ai
Google AI Studio Google’s browser workspace for testing Gemini models, building prompts and exporting application code.aistudio.google.com
Replicate Run thousands of open-weight image, video, audio and language models through one API, with prices set by model: some charge for compute time, others for inputs or outputs.replicate.comOpen weights & local
Hugging Face The home of open-weight AI: over a million models, datasets and demos, plus the libraries most of the field builds on.huggingface.co
Ollama Run open-weight models on your own machine with one command. The standard route to local Llama, Qwen, DeepSeek and Gemma.ollama.com
LM Studio A desktop app for downloading and chatting with local models, with a built-in server that mimics the OpenAI API.lmstudio.aiBenchmarks & leaderboards
Artificial Analysis Independent benchmarks across intelligence, speed and price for every serious model. The comparison tables our own model pages cite.artificialanalysis.ai
Arena Model rankings from Arena’s human evaluations, with separate boards for agents, text, coding, image and video. Read the board’s methodology and task definitions alongside its results.arena.ai/leaderboard
SWE-bench The benchmark for agentic coding: can a model resolve real GitHub issues in real repositories? The scoreboard every coding-agent launch quotes.swebench.com
ARC Prize The ARC-AGI reasoning benchmark and its prize: novel puzzles that stay hard for models while staying easy for people.arcprize.org
Epoch AI Research and datasets on AI's trajectory: training compute, hardware, data and costs, all documented and downloadable.epoch.aiResearch
arXiv The preprint server where AI research lands first, months before any conference. cs.AI, cs.LG and cs.CL are the desks to watch.arxiv.org
Hugging Face Papers Trending AI research papers, ranked by the research community's upvotes each day, week and month, each linked to its arXiv page and, where the authors publish one, its code on GitHub.huggingface.co/papers/trendingDocs & standards
OpenAI docs Official OpenAI API documentation, model catalogue, pricing and guides for text, images, audio and tools. The catalogue includes GPT-6.1 Sol and GPT-6 Astra.developers.openai.com/api/docs/overview
Claude docs Anthropic's developer documentation for the Claude models and Claude Code, including the prompting guides the lab itself maintains.platform.claude.com/docs/en/home
Model Context Protocol The open standard that lets agents reach tools and data. Required reading for anything agentic.modelcontextprotocol.ioSafety & policy
AI Incident Database A public log of real-world AI failures and harms, indexed and searchable back to the 1980s.incidentdatabase.ai
Stanford AI Index The annual report measuring the state of AI: investment, capability, adoption and regulation, with every chart downloadable.hai.stanford.edu/ai-index
OECD.AI The OECD's live observatory of AI policy: national strategies, regulation and incidents across more than 70 jurisdictions.oecd.ai
AI Security Institute The UK government institute that evaluates frontier models for dangerous capabilities and publishes what it finds.aisi.gov.ukLearn & courses
Claude Academy Anthropic's own free courses and guides: the API, Claude Code, MCP and prompting, taught by the people who built them, with certificates.academy.claude.com
DeepLearning.AI Andrew Ng's course platform and the short-course library the field learns from, most of it free to audit.deeplearning.ai
NVIDIA training NVIDIA's technical courses on the hardware side of AI: inference, CUDA, deployment and the accelerated-computing stack.nvidia.com/en-us/training