AI news: Claude Haiku 5.5, GPT-6 with interactive answers, Mistral Large 4

The thread running through this week: AI is getting much cheaper for routine work, and the choice of vendors is getting wider. Anthropic cut the price of its small model sharply, OpenAI brought a new model to every ChatGPT user, and Mistral offers a serious European alternative.

Anthropic: Claude Haiku 5.5 for high-volume work

On 7 October Anthropic released Claude Haiku 5.5, a model built for tasks that run thousands of times: summaries, classification, database queries, customer support and browser automation. For requests up to 100,000 tokens it costs USD 0.10 per million input and USD 0.50 per million output tokens. Longer requests cost USD 0.50 and USD 2.50. Anthropic says that is 90% cheaper than Haiku 4.5 for shorter requests and 50% cheaper for longer ones, or 75% less on average. It has a one-million-token context window and is available through the API, AWS, Google Cloud and Microsoft Azure.

The same day Anthropic:

• halved the price of prompt cache reads on Claude Sonnet 5.5, from USD 0.20 to USD 0.10 per million tokens;
• added monthly API credits to the Max and Team plans.

A note for developers: code written for Haiku 4.5 can break on the new model, so test before you switch.

Claude for Startups: a free year of Claude Team

On 6 October Anthropic expanded its startup programme. Companies founded in the last five years, or funded in the last two, can get a year of Claude Team for up to five Premium seats plus USD 1,000 in API credits. Venture funding is not required; bootstrapped companies can apply too. The credits expire six months after they are granted.

OpenAI: GPT-6 and interactive answers in ChatGPT

On 7 October OpenAI began rolling out GPT-6 in ChatGPT together with a feature called Intelligent UI. Answers can now include charts, buttons, forms and small calculators directly in the conversation. Plus, Pro, Business and Enterprise users got GPT-6 Sol straight away; Free and Go users get GPT-6 Luna from 8 October. Users can adjust how often visual answers appear. OpenAI itself says the model's design judgment and the range of what it can build still need work.

Mistral Large 4: a European trillion-parameter model

On 6 October French company Mistral launched Mistral Large 4 as a public preview in Mistral Studio. The model has one trillion parameters, 52 billion of them active, and covers more than 160 languages, including every official EU language. It was trained in Mistral's own data centres in Europe. Pricing is USD 1.36 per million input and USD 4.18 per million output tokens. Mistral also announced a European deployment that it runs itself under European law, and plans to release the model weights by the end of October.

What it means for your business

1. High-volume tasks now cost almost nothing. Sorting email, summarising support tickets or extracting data from invoices costs a fraction of a cent per document with a model like Haiku 5.5. If you postponed that kind of automation because of cost, run the numbers again now.
2. Check your integrations before switching models. New models count tokens differently and change the response format. Test on real data before you move a process that already works.
3. For sensitive data, choose by location too. If you serve EU clients, Mistral Large 4 with European hosting is a serious option alongside US vendors.

If you want to find out which part of your work is worth automating, see our AI automation service or get in touch.