Claude 4.5 Sonnet
Claude 4.5 Sonnet is Anthropic’s state-of-the-art coding and tool-use model, leading the SWE-bench coding benchmark and powering Claude Code, Claude.ai, and Claude in the Anthropic API.
Released
May 8, 2026
Type
multimodal
License
proprietary
Context
1,000,000 tokens
Pricing
Input
$3.00 / 1M tokens
Output
$15.00 / 1M tokens
Benchmarks
| SWE-bench Verified | 73.5% |
| MMLU | 90.1% |
| GPQA Diamond | 79.8% |
Capabilities
Links
Claude 4.5 Sonnet in the news
TechCrunch · Jun 16, 2026
Anthropic’s latest feud with the Trump admin may actually help it, sales data suggests
Anthropic's AI model, Claude, has seen a significant increase in adoption among business users, according to sales data from Ramp. This growth suggests that recent public disagreements between Anthropic and the Trump administration may not negatively impact the company's market position, and could potentially even enhance its appeal to certain customer segments. The data indicates a rising trend in businesses utilizing Claude for various tasks, implying a strong market reception for Anthropic's technology. This sustained demand from the commercial sector points to a robust business strategy that is proving effective despite external controversies. The company's focus on enterprise solutions appears to be resonating with clients seeking advanced AI capabilities.
Ars Technica · Jun 16, 2026
Anthropic "pauses" token-based billing for its Claude Agent SDK
Anthropic paused token-based billing changes for its Claude Agent SDK on June 15, reversing a decision that would have significantly increased costs for users of its automation-focused tools. The company had announced on May 13 that usage of the Claude Agent SDK, including through third-party applications and the "claude -p" command, would be billed separately from standard Claude chat interface usage. Under the original plan, SDK usage would have been charged at prevailing API rates, with subscribers receiving a monthly credit equivalent to their subscription price. This contrasted with the existing model where SDK usage was capped by the weekly limits of a user's Claude subscription tier. This existing structure allowed power users to obtain substantially more value from their subscriptions compared to paying API fees. An analysis indicated that Claude Opus subscribers could begin saving money after only two to three daily messages, with their subscriptions potentially yielding multiples of their monthly cost in API usage value.
Decrypt · Jun 16, 2026
ChatGPT Is Still Huge, But Rival AI Chatbots Are Catching Up Fast
ChatGPT's audience share fell below 50% for the first time in early 2026, according to Sensor Tower's State of AI 2026 report. This marks a significant shift from its previous dominance in the AI chatbot market. The decline is attributed to increased competition from rivals like Google's Gemini and Anthropic's Claude. Gemini has benefited from its integration as a default app on Android devices, providing a consistent user base. Claude, meanwhile, has seen a surge in adoption, partly fueled by its use within the Pentagon, indicating a growing trust and integration in governmental sectors. These developments suggest a maturing AI chatbot landscape where user acquisition and retention strategies are becoming increasingly critical for sustained growth.
Financial Times · Jun 16, 2026
US and Europe discuss access to AI models after Anthropic dispute
The United States and European Union held discussions this week regarding a potential "trusted partner" scheme that would grant European allies access to cutting-edge artificial intelligence models. This initiative emerged following a dispute involving Anthropic, a prominent AI company, and its European partners. The proposed framework aims to facilitate secure testing and evaluation of advanced AI technologies by trusted international collaborators, fostering greater transparency and cooperation in AI development. The discussions focused on establishing robust security protocols and data governance measures to ensure responsible access to sensitive AI models. Both sides acknowledged the importance of collaborative efforts in shaping the future of AI and addressing potential risks associated with its rapid advancement. The "trusted partner" concept is intended to build confidence and enable a more coordinated approach to AI governance between key global allies.
Fast Company · Jun 16, 2026
A sobering new sign for OpenAI as ChatGPT competitors gain ground
ChatGPT's share of the global AI assistant market has fallen below 50% for the first time, reaching 46% by May, according to Sensor Tower's "State of AI 2026" report. This marks a significant shift since ChatGPT's initial release in November 2022. The data, which measures usage across desktop, mobile apps, and the mobile web in 25 markets, shows Google's Gemini holding 28% market share and Anthropic's Claude at 10%. In the United States specifically, Claude's market share increased from 5% in December 2025 to 14% in May. Sensor Tower suggests that user concerns regarding OpenAI's deal with the U.S. Department of Defense may have contributed to this decline. While OpenAI stated its deal with the Pentagon includes similar "red lines" to Anthropic's previous agreement, it permits the use of ChatGPT for "all lawful purposes," a broader scope that reportedly led some users to switch to Claude. Despite the market share decrease, ChatGPT achieved a milestone in May, becoming the fastest mobile app to reach one billion users, with a total of 1.11 billion users worldwide across all platforms, still surpassing Gemini's 662 million and Claude's 245 million users.