The most important from the AI field.

OpenAI has begun rolling out o3 to all ChatGPT Plus subscribers globally — at no extra cost beyond the standard subscription of 349 kronor per month. The 80% price cut for API access makes o3 available to an entirely new category of developers and companies that previously couldn't afford the model's capabilities.

o3 is OpenAI's strongest reasoning model and performs markedly better than GPT-4o on complex multi-step tasks, mathematics and coding. For users it means a model that previously required separate API costs is now available directly inside the familiar ChatGPT interface.

The practical advice: test o3 on tasks where you normally have to follow up on the AI's answer multiple times — complex calculations, legal summaries or code with many dependencies. That's where the difference from earlier models is most visible.

Google's Gemini 2.5 Pro now takes the top spot on several coding benchmarks, including HumanEval and SWE-bench, and the margin over Claude 3.7 Sonnet is measurable. The model handles long code contexts exceptionally well and is fast at refactoring large codebases.

Tests on formal business writing show a clear weakness, however: Gemini 2.5 Pro has a tendency to mix in awkward phrasing in otherwise polished text, and idiomatic precision in formal business documents is noticeably lower than Claude's. For a business owner who wants AI that helps with customer communication, proposals and internal materials, that's the difference that decides.

The conclusion: if your primary use is code — and the code is in English — Gemini 2.5 Pro is worth testing. But if you're running mixed usage with formal prose as the core case, Claude still holds the lead where it matters for business contexts.

Anthropic is launching a new enterprise API tier called "Claude for Work" aimed specifically at mid-size and large companies that want to build internal automation workflows. The tier offers increased rate limits, dedicated capacity and a new Organizations management system that lets IT departments control access per team and project.

What sets it apart from existing API offerings is the built-in auditing log — every call is logged with timestamps and user IDs, making it easier to meet EU AI Act requirements for traceability in systems that make or support decisions. For businesses handling customer data, that's a concrete relief.

The pricing model is volume-based with a base price around $500 per month. For a company running hundreds of internal automation calls per day, that's reasonable — but it also means the threshold is too high for individual consultants and micro-businesses. However, it does open up for resellers and system integrators to build Claude-powered products for the market.