Open-source AI models put more pressure on OpenAI and Anthropic

In short: OpenAI and Anthropic face growing competition from open-source AI as they move closer to possible stock market listings.

What's going on

OpenAI and Anthropic are two of the best-known companies building large AI models. They are also widely seen as future candidates for IPOs, which means selling shares to the public for the first time. At the same time, competition is getting tougher, especially from open-source AI models (models shared publicly so others can use and modify them, like a free recipe anyone can adapt).

The Financial Times points to new open-source models that are starting to match the best paid models on popular tests. One example is GLM 5.2 from Chinese AI lab Zhipu, which the article says has reached similar results to leading models on widely followed benchmarks. The model has also been praised as a “coding agent”, meaning it can help write and edit software code by following steps, like an assistant that can take a task list and work through it.

Cost is another big factor. Many companies are seeing “sticker shock” from rising AI bills, which makes cheaper models more appealing for simpler work. The article also notes that new middle-layer tools are emerging that can split one job across several AI models, which could make it easier for customers to switch away from any single provider.

What to watch

For OpenAI and Anthropic, the key question is whether they can still charge a premium by offering features customers cannot easily get elsewhere. Investors may also watch how these companies handle concerns about misuse, regulation, and claims that some rivals copy their models without permission. The more customers feel they can swap one model for another, the harder it may be for any one AI company to stand out.

Source: Financial Times