Anthropic has released Claude Opus 4.6 as the San Francisco-based firm reportedly prepares for a potential 2026 listing that could value it north of US$350 billion.
The model outperforms OpenAI's GPT-5.2 and Google's Gemini 3 Pro across multiple benchmarks, particularly in agentic coding and knowledge work applications.
On GDPval-AA - measuring performance on economically valuable tasks in finance and legal domains - Opus 4.6 beats GPT-5.2 by 144 Elo points, whilst also topping Terminal-Bench 2.0 for agentic coding.
The system leads on BrowseComp for locating hard-to-find information online, demonstrating superior retrieval capabilities across complex search tasks.
Opus 4.6 is available at $5 per million input tokens and $25 per million output tokens, maintaining pricing parity with the previous flagship model.
Vibe working?
Anthropic's head of product for enterprise, Scott White, said the release signals a transition into "vibe working" - where AI systems move beyond executing specific commands to understanding broader business context and strategic intent.
The concept builds on vibe coding, a term former OpenAI co-founder Andrej Karpathy coined in February 2025 to describe using AI to generate code through natural language prompts.
In vibe coding, developers describe projects to large language models, which generate source code based on the prompt, whilst focusing on iterative experimentation rather than rigid correctness.
White's vibe working extends this approach across knowledge work generally, allowing workers to provide high-level direction whilst the system determines implementation pathways autonomously.
Opus 4.6 features a 1 million token context window - five times larger than its predecessor - allowing it to maintain coherence across massive documents and sprawling codebases without context rot.
The system includes adaptive thinking, where the model determines when to apply extended reasoning versus quick responses based on task complexity.
Enterprise focus
Anthropic draws roughly 80% of its revenue from corporate customers, positioning itself firmly in enterprise territory rather than consumer markets.
The company reported over 300,000 business customers as of early 2026, with the majority operating on Team and Enterprise tier subscriptions.
Software stocks cratered following Anthropic's Friday release of industry-specific plugins for its Cowork tool, with Thomson Reuters dropping 15.83% on Tuesday as investors priced in displacement risk.
LegalZoom fell nearly 20% the same session, reflecting broader market anxiety about AI substitution in specialised research and legal tech verticals.
Opus 4.6 can split coding duties across teams of agents rather than relying on single-threaded execution, whilst also integrating with PowerPoint and Excel for productivity workflows.
AI agents like OpenClaw
OpenClaw (formerly Clawdbot) is a free and open-source autonomous AI agent created by Austrian developer Peter Steinberger that executes tasks locally on user hardware.
The agent integrates with WhatsApp, Telegram and Discord, giving users interface options across messaging platforms whilst maintaining local processing for privacy.
Users report it autonomously managing insurance claims, booking flights and conducting market research without human intervention, though security experts have flagged potential attack vectors.
Cybersecurity firm Palo Alto Networks warned that AI agents present risks from access to private data, exposure to untrusted content and the ability to perform external communications whilst retaining memory across sessions.
IPO rumours
Multiple reports suggest Anthropic has hired law firm Wilson Sonsini to prepare for a possible IPO as early as 2026, though the company hasn't committed to timing publicly.
The startup has held preliminary discussions with major investment banks about a potential listing, though talks remain informal and underwriter selection hasn't begun.
The firm is negotiating a private funding round that could push valuation above $300 billion, nearly doubling its previous mark from late 2025.
With Anthropic estimating annualised revenue could reach roughly $26 billion by the end of 2026 - up from around $1 billion at the start of 2025 - it comes with IPO-ready financials.
Betting platform Kalshi shows Anthropic with a 72% chance of listing before OpenAI, reflecting market sentiment on relative IPO readiness between the two rivals.
