Anthropic

Anthropic Unveils Claude Sonnet 4.5, Its Most Advanced AI Model for Software Development

September 29, 2025 - Anthropic has introduced Claude Sonnet 4.5, the latest addition to its line of frontier AI models, boasting top-tier performance in software development tasks. The company claims this version is capable of delivering "production-ready" applications, rather than mere prototypes, marking a notable upgrade in dependability over previous models.

Claude Sonnet 4.5 is accessible via the Claude chatbot and Claude API, with pricing unchanged from Sonnet 4: $3 per million input tokens and $15 per million output tokens. One million input tokens equals roughly 750,000 words—more than the word count of the entire "Lord of the Rings" trilogy.

Over the past year, Anthropic's AI has gained traction among developers and enterprises, thanks to its strong performance in coding tasks. Major tech firms like Apple and Meta are reportedly using Claude models internally. Anthropic has also seen commercial success providing API access to platforms such as Cursor, Windsurf, and Replit. However, the rise of OpenAI's GPT-5 has presented stiff competition, outperforming Claude on multiple coding benchmarks.

Anthropic asserts that Claude Sonnet 4.5 leads the industry on various coding benchmarks, including SWE-Bench Verified. However, David Hershey, an AI researcher at Anthropic, told TechCrunch that benchmark results alone do not fully capture the model's capabilities.

Hershey shared that in early enterprise trials, Claude Sonnet 4.5 was able to autonomously code for up to 30 hours. During this period, the AI built an application, deployed database services, purchased domain names, and even conducted a SOC 2 audit to ensure security compliance.

Cursor CEO Michael Truell described Claude Sonnet 4.5 as "state-of-the-art coding performance" for long-term projects. Windsurf CEO Jeff Wang added that it represents a "new generation of coding models."

Anthropic also notes that Claude Sonnet 4.5 is its most aligned AI model to date, demonstrating reduced levels of sycophancy and deception. The model is also reportedly more resistant to prompt injection attacks.

Alongside Claude Sonnet 4.5, the company is rolling out the Claude Agent SDK. This toolkit is based on the same infrastructure that powers Claude Code and enables developers to create custom AI agents.

In addition, Anthropic is offering a temporary research preview titled "Imagine with Claude" for Claude Max subscribers. This feature showcases the AI generating code in real-time, without predefined scripts or static functionality.

Claude Sonnet 4.5's release comes just under two months after the debut of Claude Opus 4.1, underscoring the rapid pace of AI development cycles. In today's competitive landscape, companies are frequently launching new models in quick succession, making sustained leadership increasingly difficult.

Source: https://techcrunch.com/2025/09/29/anthropic-launches-claude-sonnet-4-5-its-best-ai-model-for-coding/

Anthropic Report Highlights Uneven AI Adoption Across Regions and Enterprises

September 16, 2025 – Anthropic has released the third edition of its Economic Index, revealing sharp disparities in how AI, particularly its Claude.ai platform, is being adopted across geographies and enterprises.

According to the report, AI adoption in the U.S. has doubled over the past two years, with 40% of workers now using AI tools at work, up from 20% in 2023. The analysis, based on one million Claude.ai conversations and API data, shows that while coding tasks still dominate usage at 36%, educational and scientific uses are rising steadily.

Enterprise usage of Claude’s API reveals a strong trend toward automation: 77% of business use cases are fully delegated tasks, compared to a 50/50 split between automation and augmentation among general Claude.ai users. Businesses also prioritize higher-cost, high-value tasks, showing low price sensitivity when deploying AI programmatically.

Geographic data paints a stark picture of global inequality in AI use. High-income nations like Israel, Singapore, and Canada show significantly higher usage per capita—up to 7x more than expected—while lower-income regions such as India (0.27x) and Nigeria (0.2x) lag far behind.

Within the U.S., Washington D.C. leads per-capita adoption, followed closely by Utah, both surpassing tech-heavy California. High-usage states show a broader mix of tasks, including education and document editing, while lower-usage regions remain focused on software development.

Anthropic warns that this uneven adoption could deepen existing economic divides, with AI’s productivity gains concentrated in already prosperous areas. The company has open-sourced its data to support further research and informed policymaking.

Source: https://www.anthropic.com/research/anthropic-economic-index-september-2025-report

Anthropic Faces Outage Affecting Claude, APIs, and Console

September 10, 2025 - Anthropic experienced a service disruption earlier today, temporarily affecting its APIs, Console, and Claude AI platform.

Reports of the outage began circulating on GitHub and Hacker News around 12:20 p.m. ET. Just eight minutes later, Anthropic acknowledged the issue with a status update confirming the downtime across Claude, the Console interface, and API services. At the time of reporting, the company stated that multiple fixes had been applied and that it was actively monitoring the situation.

“We’re aware of a very brief outage of our API today shortly before 9:30 a.m. PT,” an Anthropic spokesperson told TechCrunch. “Service was quickly restored.”

This is not the first time Anthropic has faced technical setbacks. The company has encountered several glitches in recent months, particularly with Claude and its underlying models.

During the outage, users took to online forums to express both frustration and humor. One GitHub user joked that the software engineering community was left "twiddling its thumbs," while a Hacker News participant referenced a prior incident with a quip: “Nooooo I’m going to have to use my brain again and write 100% of my code like a caveman from December 2024.”

Source: https://techcrunch.com/2025/09/10/anthropic-reports-outages-claude-and-console-impacted/

Anthropic Secures $13B Series F Round at $183B Valuation

September 2, 2025 - AI company Anthropic has raised $13 billion in a Series F funding round, pushing its post-money valuation to $183 billion. The funding will be directed toward expanding enterprise adoption, advancing AI safety research, and fueling international growth.

The round was co-led by Iconiq, Fidelity Management & Research Company, and Lightspeed Venture Partners, according to Anthropic’s blog post. Other participants included a broad range of institutional investors, venture capital firms, sovereign wealth funds, private equity, and asset managers—among them Altimeter, Baillie Gifford, BlackRock, Blackstone, Coatue, D1 Capital Partners, Insight Partners, Ontario Teachers’ Pension Plan, and the Qatar Investment Authority.

“We are seeing exponential growth in demand across our entire customer base,” said Krishna Rao, CFO of Anthropic. “This financing demonstrates investors’ extraordinary confidence in our financial performance and the strength of their collaboration with us to continue fueling our unprecedented growth.”

This funding follows a $3.5 billion round in March 2025, which valued the company at $61.5 billion post-money.

The latest raise comes amid reports that Anthropic was in talks to secure $3 billion to $5 billion in funding at a $170 billion valuation. The company has seen notable growth in 2025, with annual recurring revenue surging from $1 billion to $5 billion, driven by increased API usage and growing enterprise demand.

“Anthropic now serves over 300,000 business customers, and our number of large accounts—customers that each represent over $100,000 in run-rate revenue—has grown nearly 7x in the past year,” the blog post added.

Claude Code, a tool favored by developers, has also been a significant growth driver. The product now generates over $500 million in run-rate revenue, with usage spiking more than tenfold over the last three months.

Despite this momentum, competing with firms like OpenAI and Cursor continues to demand substantial capital. CEO Dario Amodei, in a memo cited by Wired, acknowledged the complexities of funding, stating he wasn’t “thrilled” about accepting money from sovereign wealth funds tied to authoritarian regimes but admitted it’s difficult to exclude “bad people” from investing while running a business.

Source: https://techcrunch.com/2025/09/02/anthropic-raises-13b-series-f-at-183b-valuation/

Anthropic Introduces “Claude for Chrome” as a Browser-Based AI Agent

August 26, 2025 - Anthropic has introduced a research preview of a browser-integrated AI agent, Claude for Chrome, powered by its Claude AI models. The new tool is initially being rolled out to a group of 1,000 users subscribed to Anthropic’s Max plan, which is priced between $100 and $200 per month. A waitlist is also open for other interested users.

By installing a Chrome extension, selected users can now interact with Claude via a side panel that retains context of browser activity. The AI assistant can also be granted permissions to perform certain actions within the browser, enabling it to help complete tasks on behalf of the user.

The browser is fast becoming a competitive space for AI integration, as tech companies race to bring AI tools directly into users’ web experiences. Perplexity recently introduced its own browser called Comet, featuring an AI agent capable of handling various tasks. OpenAI is reportedly close to launching its own AI-powered browser with features comparable to Comet. Google, meanwhile, has rolled out Gemini integrations within Chrome in recent months.

The urgency surrounding AI-powered browsers is further heightened by an ongoing antitrust case involving Google. A decision is expected soon, with the federal judge in the case suggesting Google might be compelled to divest Chrome. Perplexity has already made an unsolicited $34.5 billion offer for the browser, and OpenAI CEO Sam Altman has expressed potential interest in acquiring it.

Anthropic acknowledged in a blog post that giving AI agents browser access introduces new safety challenges. Last week, Brave’s security team identified a vulnerability in Comet that made it susceptible to indirect prompt-injection attacks, where concealed code on a webpage could deceive the agent into executing harmful instructions.

Perplexity has since addressed the issue, according to Jesse Dwyer, the company’s head of communications.

Anthropic said the Claude Chrome agent is designed to help identify and mitigate such safety threats. The company has implemented safeguards against prompt injection attacks, which it says have reduced successful attacks from 23.6% to 11.2%.

Users can adjust settings to restrict Claude’s access to specific sites, and the tool is pre-configured to block domains offering financial services, adult material, or pirated content. Claude will also request user approval before taking any high-risk actions, such as publishing content, making purchases, or sharing personal data.

This isn’t Anthropic’s first attempt at AI agents with system control. In October 2024, the company launched an agent capable of operating a PC, though it struggled with speed and reliability at the time.

Since then, AI agents have improved significantly in handling basic tasks. According to TechCrunch, tools like Comet and ChatGPT Agent now perform simple browser-based actions reliably, though many still face difficulties with more complex operations.

Source: https://techcrunch.com/2025/08/26/anthropic-launches-a-claude-ai-agent-that-lives-in-chrome/

Anthropic Expands Claude Code Access with Enterprise Integration

August 20, 2025 - Anthropic has unveiled a new subscription plan that brings Claude Code into its Claude for Enterprise suite. Previously accessible only through individual accounts, the command-line programming tool is now part of a larger enterprise package, enabling more robust integrations and enhanced administrative capabilities.

"This is the most requested feature from our business team and enterprise customers," Anthropic product lead Scott White told TechCrunch.

This move strengthens Anthropic's position against competitors like Google and GitHub, both of which launched their command-line tools with enterprise compatibility.

Since launching in June, Claude Code has gained traction as a popular command-line development tool, known for its agentic functionality that diverges from traditional IDE-based solutions. However, the tool's rising popularity has also surfaced challenges, particularly among individual users who faced unforeseen usage limits. The new enterprise package aims to address these concerns by offering scalable spending controls for high-volume users.

Anthropic sees strong potential in the synergy between Claude Code and its Claude.ai chatbot. Under the new plan, enterprises can seamlessly manage both tools, integrating Claude Code with internal data systems and developing prompts collaboratively with the Claude chatbot.

White highlighted transformative use cases, especially around customer feedback integration. By combining Claude.ai with internal tools, businesses have been able to efficiently synthesize large volumes of customer feedback into actionable product strategies.

"There's something magical about blending customer feedback, getting the voice of your customer and then helping to think about solutions that you might be able to prototype and build that address their unique challenges," White said. "It's something that as a product manager was simply not possible for me even a year ago."

Source: https://techcrunch.com/2025/08/20/anthropic-bundles-claude-code-into-enterprise-plans/

Anthropic Acquires Humanloop Team to Boost Enterprise AI Capabilities

August 13, 2025 - Anthropic has onboarded the co-founders and most of the core team behind Humanloop, a startup focused on prompt management, LLM evaluation, and observability, in a strategic move to enhance its enterprise offerings.

Although financial terms were not disclosed, the transaction appears to follow an acqui-hire model, reflecting the intensifying competition for AI expertise. Humanloop’s leadership—CEO Raza Habib, CTO Peter Hayes, and CPO Jordan Burgess—have all joined Anthropic, accompanied by about a dozen of the company's engineers and researchers.

Anthropic, recognized for its advances in agentic and coding capabilities, is rapidly expanding in the enterprise sector. While an Anthropic spokesperson clarified that the company did not purchase Humanloop’s intellectual property or other assets, the value lies in the team's knowledge and experience. Humanloop has built tools enabling enterprises to deploy safe and reliable AI solutions at scale.

Brad Abrams, API product lead at Anthropic, stated, “Their proven experience in AI tooling and evaluation will be invaluable as we continue to advance our work in AI safety and building useful AI systems.”

As competition tightens, particularly with OpenAI and Google DeepMind, enhancing its AI tooling could give Anthropic an edge in enterprise readiness and overall performance.

Founded in 2020 as a University College London spinout, Humanloop participated in Y Combinator and the Fuse Incubator before raising $7.91 million in seed funding, led by YC and Index Ventures, according to PitchBook. The company earned recognition for supporting clients like Duolingo, Gusto, and Vanta in building and fine-tuning AI applications.

Last month, Humanloop informed its users of an impending shutdown in anticipation of an acquisition.

This hire aligns with Anthropic’s strategy, especially as it recently introduced extended context windows for enterprise customers, enhancing model capabilities. Earlier this week, Anthropic secured a deal with the U.S. government’s central purchasing body, offering AI services across all branches for just $1 per agency in the first year—a direct challenge to OpenAI’s similar pricing.

The addition of Humanloop’s team supports Anthropic’s positioning as a "safety-first" AI firm. Humanloop’s frameworks for evaluation, safety, and bias mitigation align with Anthropic's core mission.

“From our earliest days, we’ve been focused on creating tools that help developers build AI applications safely and effectively,” said Raza Habib, former Humanloop CEO. “Anthropic’s commitment to AI safety research and responsible AI development perfectly aligns with our vision.”

Source: https://techcrunch.com/2025/08/13/anthropic-nabs-humanloop-team-as-competition-for-enterprise-ai-talent-heats-up/

Anthropic Surpasses OpenAI as Top AI Provider Among Enterprises

July 31, 2025 - AI research firm Anthropic has overtaken OpenAI to become the leading AI model provider for enterprises, according to a report released Thursday by Menlo Ventures.

Anthropic currently commands 32% of the enterprise large language model (LLM) market by usage, with OpenAI trailing at 25%. This marks a significant shift from 2023, when OpenAI held 50% of the market and Anthropic just 12%.

The report highlights a steady decline in OpenAI’s enterprise usage over the past two years, as Anthropic's share has grown. Google has also seen a rise in enterprise adoption of its models, though specific figures were not disclosed.

In the enterprise coding segment, Anthropic’s lead is even more pronounced. The company now holds 42% of the market, double OpenAI’s 21%, making it the dominant provider in that category.

The launch of Anthropic’s Claude 3.5 Sonnet in June 2024 was cited as a turning point in the company’s growth. The subsequent release of Claude 3.7 Sonnet in February 2025 further accelerated its adoption among enterprise users.

These findings confirm industry speculation that enterprises and startups increasingly favor Claude over OpenAI’s ChatGPT for development tasks. Despite this, OpenAI continues to dominate consumer usage, with over 2.5 billion daily prompts sent to ChatGPT, as reported last week.

The Menlo Ventures study also notes that enterprises overwhelmingly prefer closed models. Both Anthropic and OpenAI operate on closed-source frameworks. More than 50% of enterprises surveyed said they do not use open source models. In fact, only 13% of enterprise daily workloads involved open source models by mid-2025, a drop from 19% earlier in the year. Meta remains the leader in the open source LLM market.

Source: https://techcrunch.com/2025/07/31/enterprises-prefer-anthropics-ai-models-over-anyone-elses-including-openais/

Anthropic Hits $3 Billion in Annualized Revenue Amid Rising AI Demand

May 31, 2025 - AI developer Anthropic has reached an annualized revenue run rate of $3 billion, according to two sources familiar with the company’s finances, marking a significant milestone for the generative AI sector in enterprise adoption.

The company’s projected revenue, calculated by extrapolating current sales over a year, has tripled since December 2024, when it stood at nearly $1 billion. Anthropic crossed the $2 billion threshold around late March and reached $3 billion by the end of May, one source said.

While OpenAI’s ChatGPT has been embraced by consumers, many enterprises are still in the experimentation phase with AI. Anthropic’s revenue growth reflects increasing demand from businesses for AI models delivered as a service, according to one of the sources.

Code Generation Drives Growth

A major factor behind Anthropic’s surge is its success in code generation. The San Francisco-based company, backed by Alphabet (GOOGL.O) and Amazon.com (AMZN.O), is renowned for AI models that excel in computer programming. Adoption of products in the codegen space, often powered by Anthropic’s technology, has accelerated significantly in recent months.

This demand positions Anthropic as a standout among software-as-a-service (SaaS) vendors. The company’s rapid revenue climb is unprecedented, according to Meritech General Partner Alex Clayton, who said, “We’ve looked at the IPOs of over 200 public software companies, and this growth rate has never happened.” Clayton is not an Anthropic investor and has no inside knowledge of its sales.

By comparison, Snowflake took six quarters to move from $1 billion to $2 billion in run-rate revenue, Clayton noted. Anthropic’s consumer-facing Claude chatbot also contributes to its revenue, though its adoption trails OpenAI’s ChatGPT. In April, Claude’s web traffic was about 2% of ChatGPT’s, according to Similarweb.

Rivalry With OpenAI Intensifies

OpenAI, Anthropic’s chief competitor, is projected to close 2025 with over $12 billion in total revenue, up from $3.7 billion in 2024, three people familiar with the matter said. Unlike Anthropic’s run-rate metric, OpenAI’s figure represents total annual revenue.

The two companies appear to be diverging in focus. OpenAI is becoming more consumer-centric, with most revenue from ChatGPT subscriptions, while Anthropic is gaining traction in enterprise AI. OpenAI’s enterprise customer base has also grown, with ChatGPT Enterprise paying seats increasing to 3 million in May from 2 million in February. Notable clients include T-Mobile (TMUS.O) and Morgan Stanley (MS.N).

Founded in 2021 by ex-OpenAI employees, Anthropic raised $3.5 billion earlier this year, valuing the company at $61.4 billion. OpenAI’s current valuation stands at $300 billion.

Source: https://www.reuters.com/business/anthropic-hits-3-billion-annualized-revenue-business-demand-ai-2025-05-30/