The artificial intelligence industry’s accelerating power struggle spilled well beyond Silicon Valley this week, as Anthropic’s release of a new flagship model coincided with sharp market swings, intensifying rivalry with OpenAI and fresh questions about how quickly AI could reshape enterprise software.
On February 5, 2026, Anthropic unveiled Claude Opus 4.6, deploying it immediately across its own platforms, the Claude API, major cloud providers, and GitHub Copilot. The timing was notable: the launch came just 72 hours after OpenAI introduced a new desktop version of Codex, underscoring how tightly coupled technical advances and competitive signaling have become in the race for AI dominance.
Investors reacted first. Two days earlier, on February 3, a $285 billion selloff swept through software and services stocks. Goldman Sachs’ basket of U.S. software shares fell 6%—its steepest one-day drop since April’s tariff-driven rout—after Anthropic released a new legal automation tool that rattled incumbents. Shares of Thomson Reuters and Legalzoom.com dropped by double digits, while European information providers RELX and Wolters Kluwer were also hit.
A model designed for scale, not demos
Behind the market turbulence is a model aimed squarely at enterprise workloads. Claude Opus 4.6 introduces a one-million-token context window, allowing it to process and reason over vast bodies of text or code without the degradation known as “context rot.” Anthropic said the model scored 76% on the MRCR v2 long-context retrieval benchmark, up from 18.5% for its predecessor, Sonnet 4.5.
The company also rolled out “agent teams” within Claude Code as a research preview, enabling multiple autonomous AI agents to work in parallel on different parts of a software project—such as frontend, APIs, or migrations—while coordinating with one another. Anthropic described the feature as particularly suited to complex planning and coding tasks.
Benchmark results released alongside the launch were aggressive. Opus 4.6 posted the highest score yet on Terminal-Bench 2.0, an agentic coding evaluation, and led all frontier models on Humanity’s Last Exam, a multi-disciplinary reasoning test. On GDPval-AA, which measures performance on economically valuable tasks in finance, legal work and other professional domains, Anthropic said Opus 4.6 outperformed OpenAI’s GPT-5.2 by roughly 144 ELO points, winning about 70% of head-to-head comparisons.
The model is available immediately at unchanged prices—$5 per million input tokens and $25 per million output tokens, with premium pricing for prompts exceeding 200,000 tokens. Anthropic has advised users to lower the default “effort” setting from high to medium if the model appears to over-analyze simpler tasks.
Enterprise momentum and a widening rivalry
Anthropic’s push is backed by rapid commercial uptake. Claude Code reached a $1 billion run-rate just six months after its general availability in May 2025, the company said. Enterprise customers include Uber, Salesforce, Accenture, Spotify, Rakuten, Snowflake, Novo Nordisk and Ramp, using the tools across software engineering, finance, and trust-and-safety operations.
Those numbers have fed investor enthusiasm. Earlier this month, Anthropic signed a term sheet for a $10 billion funding round valuing the company at $350 billion. Surveys from Andreessen Horowitz show how sharply enterprise adoption has shifted: 44% of enterprises reported using Anthropic in production as of January 2026, up from near zero in March 2024. OpenAI remains the most widely used provider, with 77% of companies deploying its models in production, but Anthropic claims slightly higher production-use rates among its own customers—75% already live, with 89% either testing or deployed.
Spending is rising accordingly. Data from a16z show average enterprise LLM spending climbed to $7 million in 2025, up 180% from $2.5 million in 2024, with projections of $11.6 million for 2026.
Not everyone accepts the market’s fears. Nvidia chief executive Jensen Huang dismissed the idea that AI will wholesale replace existing software as “illogical.” JPMorgan’s Mark Murphy said it was an “illogical leap” to assume a new large-language-model plug-in could displace every layer of mission-critical enterprise systems.
Anthropic is also extending Claude into everyday productivity tools. The company released a research preview integrating Claude into PowerPoint, allowing AI-assisted presentation creation alongside documents and spreadsheets—an unusual move given Microsoft’s 27% stake in OpenAI. Anthropic said the goal was to meet users inside the tools they already use.
Safety remains a central theme. Anthropic said Opus 4.6 maintains strong alignment, shows a low rate of problematic responses, and has the lowest over-refusal rate of any recent Claude model. The company has added six new cybersecurity probes to detect misuse and is using the model itself to identify and patch vulnerabilities in open-source software.
The rivalry is now playing out publicly as well. Both Anthropic and OpenAI plan to air Super Bowl advertisements, with Anthropic’s spots mocking OpenAI’s experiments with ads in ChatGPT. OpenAI chief executive Sam Altman responded on X, calling the ads “funny” but “clearly dishonest,” and accusing Anthropic of seeking to control how people use AI while selling “an expensive product to rich people.”
The exchange captures the broader split: OpenAI is moving toward advertising to monetize a massive free user base, while Anthropic is betting almost entirely on enterprise sales and premium subscriptions. With Claude Opus 4.6, that bet has become louder—and more consequential—for markets, customers and competitors alike.
