Paradigm has reportedly raised a $1.2 billion AI fund, signaling a major expansion beyond crypto and intensifying competition for early AI deals.

Anthropic launched Claude Sonnet 5 with stronger agent features and lower pricing, aiming to make autonomous AI workflows cheaper to deploy.

8090 Labs, an AI coding startup tied to investor and operator Chamath Palihapitiya, has raised a $135 million Series A and named Palihapitiya as CEO, according to TechCrunch and other media reports. The move pairs a large early-stage financing with a hands-on leadership change at a time when competition in AI coding tools is intensifying and investors are concentrating capital around startups that claim they can turn code generation into repeatable enterprise software workflows.

Regional outlets Wamda and CairoScene report that startup 1001 has closed a $30 million Series A round focused on building sovereign AI in the Gulf. Public details are still thin, but the funding points to growing demand for GCC-based AI infrastructure, local model deployment, and tighter control over data residency as governments and enterprises reassess where their AI systems run.

OKX says it is opening a developer marketplace for AI agents that can discover services, hire one another, and settle payments with stablecoins using blockchain-based identity and reputation tools. The move extends the crypto exchange beyond trading and places it in a fast-forming market for agent infrastructure, where payments, trust, and dispute resolution remain unresolved problems.

Google says it is making the biggest change to its Search box in 25 years, expanding the core interface beyond typed keywords to accept images, files, video and follow-up conversation inside the main Search flow. The redesign matters because it collapses the distinction between classic Google Search, AI Overviews and AI Mode, pushing AI interaction into the default front door of Google’s most important product while raising new questions for publishers, advertisers and enterprise teams that depend on web traffic and search visibility.

Amazon shares moved modestly higher in overnight trading after the company denied that a reported restructuring of its commercial relationship with Anthropic would make AI services more expensive. Based on limited wire-style coverage available in this story cluster, the key development is not a new product launch but Amazon’s effort to calm concerns about the economics of one of its most important AI partnerships. For builders and enterprise buyers, the episode underscores how closely model access, cloud distribution, and strategic investment terms are now tied to perceived AI infrastructure costs.

Singapore-based Akro AI has reportedly raised a $700,000 pre-seed round, according to Tech in Asia and DealStreetAsia deal coverage. Publicly available source evidence is thin, leaving key details such as investors, product scope, and timing unclear. Even so, the deal stands out as another data point in Southeast Asia’s early-stage AI funding market, where smaller rounds increasingly matter for builders testing narrow enterprise use cases before larger capital commitments arrive.

Anthropic has introduced Cowork, a research-preview feature in the Claude macOS app that lets users give Claude access to a chosen folder so it can read, edit, and create files while carrying out multi-step tasks. The launch extends ideas from Claude Code to non-technical workflows, but it is limited to Claude Max subscribers for now and comes with explicit warnings about destructive actions, prompt injection, and the still-immature safety model for real-world AI agents.

Nous Research has released NousCoder-14B, a 14B-parameter open-source coding model that the company says reaches 67.87% on LiveCodeBench v6 after a four-day reinforcement learning run on 48 Nvidia B200 GPUs. The launch lands as developer attention is shifting toward more autonomous coding tools, raising questions about whether open models can keep pace with proprietary systems and whether reproducible training stacks will matter as much as benchmark scores.

Salesforce has launched a rebuilt Slackbot that moves beyond reminders and notifications into search, drafting and task execution inside Slack. The product is now generally available for Slack Business+ and Enterprise+ customers, with Salesforce positioning it as a central AI interface for work while it competes with Microsoft Copilot and Google’s Gemini-based workplace tools.

Omen AI has raised a $31 million Series A to expand a sensor system that monitors the health of liquid cooling fluids in data centers, aiming to catch bacterial contamination and component wear before they force expensive rack shutdowns. The startup is repositioning from heavy equipment fluid monitoring into AI infrastructure at a moment when hotter, denser GPU deployments are making cooling reliability a larger operational risk.

Arena, the Berkeley-born startup behind one of the most widely watched crowdsourced AI model leaderboards, told TechCrunch it has reached $100 million in annualized run-rate revenue roughly eight months after launching its commercial evaluation product. The milestone suggests that AI benchmarking and post-training feedback are becoming a business in their own right, not just a research exercise, though the company says its revenue is consumption-based rather than contracted recurring ARR.

A pricing backlash around Anthropic’s Claude Code is creating room for open-source rivals. VentureBeat reports growing developer frustration with Claude Code’s subscription tiers and rate limits, while Block’s Goose is gaining attention by offering a model-agnostic coding agent that can run locally, offline, and without subscription fees. The comparison is not one-to-one on model quality or ease of use, but it highlights a broader shift: AI coding tools are increasingly competing not just on capability, but on control, privacy, and cost predictability.

TechCrunch reports concerns that AI wealth is concentrating among a small group while software workers face uncertainty.