AI News
> ▌
Why Opus 4.8 Pulled Me Back to Claude
Every
R&D Part 1: Here to Win
OpenAI
Deno 2.8 Quietly Fixed 3 Huge Node Problems #nodejs #programming
Better Stack
GitHub Trending Weekly #34: workshop, forkd, HRM-Text, codegraph, ELF, ASCII-Aquarium, mailflare
Github Awesome
Claude Plans, Gemini Designs: One Workflow for Beautiful Frontends (LIVE)
Cole Medin
Can We Transfer SKILLS w/ Memory to HERMES?
Discover AI
Finally! Claude Code Saves Your Model Preferences #claude #aicoding
DIY Smart Code
Vibe Coding is Serious
The PrimeTime
The GTA CEO is so based
The PrimeTime
This One File Fixed My Dev Environment (Devbox)
Better Stack
Holy sh*t I think Anthropic is profitable now
Theo - t3․gg
Codex 4.0 (CRAZY NEW UPGRADES): THIS IS INSANITY!
AICodeKing
I was wrong about GPT 5.5
Ben Davis
Code mode in the Mistral Vibe web app
Mistral AI
Custom skills and scheduling with Mistral Vibe
Mistral AI
Investment portfolio analysis with Mistral Vibe
Mistral AI
Introducing the Mistral Vibe extension for VS Code
Mistral AI
Mistral Vibe: the agent for long-horizon work
Mistral AI
Use Connectors in Workflows
Mistral AI
Use Connectors in Vibe
Mistral AI
Claude Opus 4.8 Remains Unconfirmed
Anthropic’s official pages still show Opus 4.7 as the latest published flagship model, with no public announcement, model card, or release note for Opus 4.8.
Microsoft Plans In-House Coding Model
The Information says Microsoft plans to show a homegrown coding model at Build next week, alongside new reasoning, speech, transcription, and image models. The move looks aimed at making GitHub Copilot less dependent on OpenAI and Anthropic while tightening control over cost and performance.
Opus 4.8 Distills Mythos Into Cheaper Tier
The post frames Opus 4.8 as a distillation of Mythos, which reads like a model-compression or specialization story inside Anthropic’s Claude line. Based on current public references, this looks more like leak-driven model chatter than an official launch announcement, with the implication that Anthropic is segmenting capability tiers instead of shipping a single general-purpose upgrade.
BridgeMind hits $193K ARR via vibe coding
BridgeMind AI founder Matthew Miller reports reaching $193,248 in Annual Recurring Revenue as part of his "vibe coding" challenge. The project demonstrates the commercial viability of "agentic organizations" where small teams leverage autonomous AI agents to ship and scale production software at high velocity.
Claude Opus 4.8 surfaces in Claude Code
Claude Opus 4.8 appears in Claude Code’s model selector, gated behind Claude Desktop 1.3036.0. The string suggests Anthropic is wiring up a newer Opus variant ahead of a public release, but it is still just a source-code signal.
Microsoft debuts homegrown AI coding models
Microsoft is unveiling a suite of in-house AI models at next week's Build conference, led by a new coding model designed to power GitHub Copilot and reduce reliance on OpenAI.
Midjourney founder: diffusion wins as FLOPS outpace memory
David Holz argues that diffusion models are the superior long-term architecture because they scale with cheap compute (FLOPS) while autoregressive models remain bottlenecked by expensive memory bandwidth.
Coinbase builds read-only Temporal MCP server
Coinbase engineers developed a read-only Model Context Protocol (MCP) server that lets AI assistants debug Temporal workflows directly from code editors. The tool enables natural language troubleshooting by correlating live production state with local source code.
Claude powers Polymarket arbitrage workflows
A viral retweet frames Claude as a practical tool for trading-adjacent automation, specifically analyzing mispriced Polymarket markets to surface arbitrage opportunities. The post is less a product launch than a signal of how users are adopting Claude for high-leverage, semi-structured research tasks that combine reasoning, pattern matching, and market scanning.
CodeRabbit Draws Demo Crowds at App.js Conf
A retweeted post from CodeRabbit says the team is having a hectic time at App.js Conf and is asking for more hands because they cannot keep up with showing people the product. This reads as a traction and field-interest signal rather than a product announcement, with the main takeaway being that the booth/demo activity is pulling in more attention than the team can comfortably handle.
Anthropic hits first profit on $10.9B Q2 revenue
Anthropic is poised to record its first operating profit in Q2 2026, driven by a massive $10.9 billion revenue run and a strategic pivot to enterprise sales. The financial turnaround highlights the explosive monetization potential of developer-focused coding agents like Claude Code.
Anthropic hits profitability as Claude Code usage surges
Anthropic achieved its first operating profit in Q2 2026, driven by a massive shift toward usage-based enterprise pricing. The company's agentic CLI, Claude Code, has become its primary revenue engine by consuming high volumes of tokens for autonomous coding tasks.
Anthropic readies Opus 4.8 release amid leaks
Rumors of an imminent Claude Opus 4.8 launch swirl as model slugs appear in staging and OpenAI drops stealth updates. The anticipated release signals a pivot toward deeper agentic capabilities and integrated developer workflows.
Pocock: Fewer test seams boost agents
TypeScript authority Matt Pocock argues that minimizing test seams is the key to unlocking AI agent productivity. By focusing on "single-seam" problems like compilers and pure libraries, developers can reduce the architectural "context bounce" that often derails LLM-led refactoring and autonomous coding tasks.
ElevenLabs, Greece partner on voice AI gov services
ElevenLabs signed a Memorandum of Understanding with the Greek government to integrate voice AI into the gov.gr portal, automate public service call centers, and preserve regional dialects like Cretan. The initiative aims to modernize bureaucracy and tourism through natural language interaction and linguistic heritage preservation.
Dev lets Claude trade BTC overnight, nets $95 profit
A developer gave Claude a $20 budget to autonomously script and execute Bitcoin trades overnight, waking up to a functional trading bot and a $95 profit across five trades.
Aaronson says AI turns mathematicians into curators
Scott Aaronson says recent AI results in mathematics, including a GPT-5.5 Pro solution to Erdős’s Unit Distance Problem, suggest humans may increasingly focus on choosing questions and interpreting model outputs. He extends the argument to AI-written fiction and the Vatican’s AI encyclical as signs of a broader cultural shift.