Best of Anthropic — 2025

1
Article
Collections·21w
Anthropic Acquires Bun to Enhance AI Coding Products' Performance and Stability
Anthropic acquired Bun, the high-performance JavaScript runtime with 7 million monthly downloads and 83,000 GitHub stars. Despite $26 million in VC funding, Bun generated no revenue. The runtime will remain open source under MIT license and integrate into Anthropic's AI coding products like Claude Code and Claude Agent SDK. The acquisition provides Bun with resources to expand its team and accelerate development while supporting Anthropic's AI-driven development infrastructure, as Claude Code reaches $1 billion in run-rate revenue.
319
38
2
Article
Where's Your Ed At·41w
Anthropic Is Bleeding Out
Anthropic appears to be losing substantial money on its Claude Code product, with users burning hundreds to thousands of dollars worth of compute while paying only $20-200 monthly subscriptions. Analysis of user data suggests the company may be losing 200-3000% on each customer, creating a massive financial drain. This explains recent aggressive price increases on enterprise customers like Cursor, which had to restructure its business model after Anthropic raised API costs. The situation represents a fundamental business model crisis where AI companies are subsidizing unsustainable usage patterns.
152
23
3
Article
Reinier·41w
Claude Code UI
Claude Code UI provides a desktop and mobile interface for Anthropic's Claude Code CLI tool, enabling developers to manage AI-assisted coding projects and sessions through a graphical interface instead of command-line interactions. The tool works both locally and remotely, offering the same functionality as the CLI with improved accessibility across different platforms.
142
6
4
Article
Simon Willison·27w
Claude Skills are awesome, maybe a bigger deal than MCP
Anthropic introduced Claude Skills, a new pattern for extending LLM capabilities using Markdown files with instructions, scripts, and resources. Skills are token-efficient (loading only when needed), depend on code execution environments, and are simpler to create than MCP implementations. The system enables general computer automation beyond just coding tasks, with skills shareable as single files or folders. Skills work with other models too, potentially sparking wider adoption than the Model Context Protocol.
138
6
5
Article
Collections·39w
Anthropic Introduces Weekly Rate Limits for Claude Code Subscribers
Anthropic has implemented weekly rate limits for Claude Code subscribers starting August 28, 2025, affecting both Pro and Max tiers. The new structure operates on 5-hour rolling sessions with weekly quotas impacting less than 5% of intensive users. Most users receive 140-280 hours of Sonnet 4 and 15-35 hours of Opus 4 weekly, while Max subscribers can access higher tiers at $100-200/month. The changes have drawn criticism from developers who feel legitimate power users are penalized, though Anthropic cites system reliability and policy violation prevention as reasons for the restrictions.
124
40
6
Video
YouTube·39w
Claude Engineer is INSANE... Upgrade Your Claude Code Workflow
Two free tools can significantly enhance Claude Code workflows: Super Claude, a configuration framework that adds 18 structured commands with development personas and flags for different stages of the software development lifecycle, and a web-based GUI that enables browser access to Claude Code from any device on the same network. Super Claude provides pre-built workflows for frontend, backend, security, and architecture tasks, while the web GUI offers cross-device accessibility for remote coding sessions.
80
2
7
Article
Javarevisited·38w
Top 5 Udemy Courses to Learn Claude Code and Claude AI in 2025
Claude AI and Claude Code are emerging as powerful tools in the AI development stack, created by Anthropic with a focus on safety and natural language understanding. Claude Code enables developers to write production-ready code through conversational prompts and automate workflows with AI agents. The article curates five Udemy courses covering different aspects: from basic Claude Code usage and full-stack AI development to advanced agent building with frameworks like LangChain, CrewAI, and AutoGen. These courses cater to various skill levels and use cases, from beginners learning AI-assisted coding to experienced developers building complex autonomous agents.
75
1
8
Article
ByteByteGo·32w
How Anthropic Built a Multi-Agent Research System
Anthropic built a multi-agent research system using an orchestrator-worker pattern with a Lead Researcher agent coordinating specialized subagents and a Citation Agent for accuracy. The system outperformed single-agent setups by 90% through parallel processing and dynamic adaptation, though it consumes 15x more tokens. Key engineering principles include proper prompt design, delegation strategies, effort scaling, and parallelization. Production challenges involve managing stateful agents, debugging non-deterministic behavior, and handling deployments without breaking running tasks.
74
9
Article
Where's Your Ed At·33w
Why Everybody Is Losing Money On AI
Generative AI companies are fundamentally unprofitable, with every major player from OpenAI to Anthropic losing billions of dollars annually. The economics are broken: companies like Cursor send 100% of their revenue to model providers, Perplexity spends 164% of revenue on compute costs, and OpenAI lost $5 billion in 2024 despite being the market leader. Inference costs are increasing rather than decreasing, making it impossible for AI startups to achieve profitability even with usage-based pricing models.
64
9
10
Article
Simon Willison·48w
Highlights from the Claude 4 system prompt
Anthropic has released the system prompts for their Claude 4 models, providing insights into prompt engineering and model personalities. The prompts offer guidance on effective use, limiting hallucinations, and maintaining model safety while they detail Claude's capabilities, including its preferences for conversation styles and ensuring copyright compliance.
62
2
11
Article
Where's Your Ed At·38w
How Much Money Do OpenAI And Anthropic Actually Make?
OpenAI and Anthropic use annualized recurring revenue (ARR) metrics to inflate their apparent success, with OpenAI claiming $12 billion ARR and Anthropic $4-5 billion ARR. However, analysis of actual monthly revenues suggests OpenAI made around $3.6 billion in 2024 and $5.3 billion through July 2025, while Anthropic made approximately $1.5 billion through July 2025. ARR multiplies one month's revenue by 12, creating misleading impressions of annual performance and allowing startups to appear more successful than actual cash flow indicates.
59
12
Article
Where's Your Ed At·42w
Anthropic and OpenAI Have Begun The Subprime AI Crisis
AI companies like Anthropic and OpenAI are operating at massive losses while subsidizing their services, creating what's termed a 'subprime AI crisis.' Anthropic expects to lose $3 billion in 2025 despite $4 billion in revenue. The popular AI-powered coding tool Cursor has grown rapidly to $500 million in annual recurring revenue, but its success depends on these loss-making AI models. As AI companies eventually raise prices to achieve profitability, dependent services and startups may face unsustainable costs, potentially triggering industry-wide disruption.
55
8
13
Article
Vibe Coding·30w
Vibe coding just got an upgrade with Claude Sonnet 4.5
Anthropic released Claude Sonnet 4.5, their latest and most capable coding model that reportedly outperforms the previous Opus 4.1 version.
54
18
14
Article
Tech Lead Digest·35w
The free tier death cult
AI coding startups are facing a critical economic crisis where unlimited plans lead to massive losses. Companies like Anthropic, Cursor, and Replit are losing hundreds or thousands of dollars per power user while charging $20-200/month. One user consumed $10,000+ worth of compute on a $200 plan, forcing Anthropic to kill Claude Code's unlimited tier. The entire AI coding industry operates on negative margins, with some companies losing 300-500% on heavy users, essentially subsidizing compute costs with venture capital funding.
51
9
15
Video
Theo - t3․gg·41w
Everyone’s mad at Cursor right now
Cursor recently changed its pricing model from 500 requests per month for $20 to a credit-based system with $20 of API usage, causing user backlash due to poor communication. Many users experienced unexpected charges when the new pricing kicked in without clear warnings. The change reflects broader industry trends as AI coding tools move away from loss-leader pricing to more sustainable models that reflect actual API costs. While Cursor's value proposition remains strong, the lack of transparency in usage tracking and billing has created uncertainty among users about when they'll hit limits or incur additional charges.
51
11
16
Article
Hacker News·46w
Claude's System Prompt Changes Reveal Anthropic's Priorities
Analysis of Claude 4.0's system prompt reveals how Anthropic uses natural language instructions to program chatbot behavior. Key changes include removal of old hotfixes (now handled in training), encouragement of search functionality, expanded artifact use cases, context optimization for coding, and new cybersecurity guardrails. The 23,000-token system prompt consumes 11% of Claude's context window and demonstrates a user-driven development cycle where observed behaviors are first addressed through prompt modifications, then incorporated into model training.
48
17
Video
Awesome·20w
Claude buys Bun, and this is really bad for devs...
Anthropic acquired Bun, the JavaScript runtime built as a faster alternative to Node.js. While Bun will remain open source and MIT licensed, concerns arise about whether future development priorities will serve the JavaScript community or primarily benefit Claude's AI coding capabilities. The acquisition highlights broader tensions around AI companies claiming to replace developers while simultaneously hiring engineers and acquiring development tools. The piece questions the credibility of predictions about AI replacing software engineers within months, comparing them to historically inaccurate tech forecasts.
45
7
18
Article
WeProDev·33w
Anthropic AI courses
Anthropic has launched an educational platform offering structured courses on AI and machine learning concepts. The platform provides learning resources for developers and professionals looking to understand and work with AI technologies, particularly focusing on Anthropic's approach to AI development and safety.
44
19
Article
Collections·30w
Anthropic Releases Claude Sonnet 4.5: State-of-the-Art AI for Coding
Claude Sonnet 4.5 achieves a 77.2% score on SWE-bench, positioning it as a leading AI coding model. Available on GitHub Copilot, VS Code, and JetBrains IDEs, it features enhanced memory management for tasks up to 30 hours, improved tool orchestration, and autonomous task handling. The model integrates with Snowflake Cortex AI and Amazon Bedrock for enterprise deployment, with pricing at $3 per million input tokens and $15 per million output tokens. Safety improvements include reduced sycophancy and better resistance to prompt injection attacks.
39
2
20
Video
Theo - t3․gg·23w
Anthropic admits that MCP sucks
Anthropic published guidance showing that code execution is 98.7% more efficient than their Model Context Protocol (MCP) specification for AI agents. The article demonstrates how writing code to interact with MCP servers reduces token usage from 150,000 to 2,000 tokens by avoiding context window bloat from tool definitions and intermediate results. This approach enables on-demand tool loading, data filtering before reaching the model, and better privacy controls, though it requires secure sandboxed execution environments.
38
2
21
Article
Where's Your Ed At·27w
This Is How Much Anthropic and Cursor Spend On Amazon Web Services
Anthropic spent $2.66 billion on AWS through September 2025, exceeding its estimated $2.55 billion revenue for the same period. The company's AWS costs increased 174% from January to September 2025, consuming 88-227% of monthly revenue depending on the period. Cursor, Anthropic's largest customer, saw its AWS bills double from $6.2M to $12.6M in June 2025 after Anthropic introduced Priority Service Tiers that significantly increased costs for prompt caching. The analysis reveals that AI model providers' operational costs scale linearly with revenue, suggesting current pricing models are unsustainable without dramatic price increases that could drive away customers.
36
5
22
Video
Community Picks·44w
MCP Server Tutorial | Build your first MCP Server with TypeScript SDK
Model Context Protocol (MCP) is a standardized way to connect AI applications with external data sources. Developed by Anthropic, MCP works like a universal connector that allows AI models to access structured context from databases, APIs, calendars, and files through custom-built servers. The tutorial demonstrates building an MCP server using TypeScript that integrates Google Calendar with AI assistants like Claude or Cursor, enabling natural language queries about meetings and schedules. MCP servers communicate via standard input/output and can be reused across different AI applications, creating new opportunities for developers to build AI-enabling infrastructure.
33
3
23
Video
Fireship·20w
Anthropic just bought your favorite JS runtime...
Anthropic acquired Bun, the high-performance JavaScript/TypeScript runtime and toolkit. Created by Jared Sumner to address slow build times in the JS ecosystem, Bun has grown to 7+ million monthly downloads and 83,000+ GitHub stars. The acquisition positions Bun as infrastructure for Anthropic's AI products like Claude Code, which already uses Bun for its CLI. The team promises Bun will remain open source and MIT licensed, though concerns exist about Anthropic's open source track record.
28
1
24
Article
J-HAYER - Builders gonna build·26w
Learn Claude's new Skills
Claude introduces Agent Skills, a new capability that allows developers to package domain expertise into reusable modules instead of managing complex monolithic prompts. Skills are composable, portable across platforms, loaded on-demand to save context window space, and can include executable code for precise operations. Companies like Box are already using Skills to automate tasks like transforming unstructured files into branded presentations.
28
4
25
Video
Theo - t3․gg·32w
It's not just you (Claude did get dumber)
Anthropic's Claude models experienced significant quality degradation and reliability issues over several months, with multiple bugs affecting model intelligence going unnoticed and unacknowledged for extended periods. The company's lack of transparency, poor infrastructure reliability, and prioritization of research over user experience has created trust issues. Alternative providers like Google Vertex and AWS Bedrock offer better uptime for Claude models, and services like OpenRouter provide failover capabilities to avoid single-provider dependency.
26
7

See all Anthropic archives