Anthropic has edged past OpenAI to become the most widely used large-language-model provider in corporate settings, according to a survey published by Menlo Ventures. The venture-capital firm, which is an investor in Anthropic, polled 150 technical decision-makers at companies deploying generative-AI applications and found that 32 % of production workloads now run on Anthropic’s Claude models. OpenAI ranked second with 25 % share, followed by Google with 20 % and Meta’s Llama with 9 %. The report attributes Anthropic’s rise to the performance of its Claude Sonnet and Claude Opus models, particularly in code generation, where they hold 42 % of developer adoption—double OpenAI’s share. Menlo Ventures also cited Anthropic’s use of reinforcement learning with verifiable rewards to improve output quality. Anthropic’s enterprise momentum accompanies rapid top-line growth. The San Francisco-based startup, valued at $61.5 billion in its latest funding round, has reached an estimated $3 billion in annual recurring revenue, the report said. While price remains a factor, survey respondents indicated they are willing to pay a premium for gains in model performance, suggesting further share shifts are likely as providers roll out new releases.