AI Made Friendly HERE

[Latest as of 2026.9.23] The two AI giants engage in a surprise same-day ‘price-cut bombardment’! 5 AI news highlights including the arrival of Claude Opus 5.5 & GPT-6 Sol/Luna and the complete shutdown of Sora

In the latest AI news as of September 23, 2026, the two giants of the generative AI industry, Anthropic and OpenAI, have clashed head-on with ‘high-performance AI price cuts’ on the same day in a shocking development.

While the trend of significantly lowering usage costs while maintaining the performance of top-tier models continues, there has been a series of unmissable news regarding practical and safety aspects, such as the complete service termination of the video generation AI ‘Sora’ and troubles surrounding the autonomous behavior of AI.

Here is an easy-to-understand explanation of the top 5 most important AI news stories you need to know today!

🥇 1st Place: Anthropic announces ‘Claude Opus 5.5’! 40% price reduction & 30% speed improvement

On September 22, 2026, Anthropic officially announced **’Claude Opus 5.5’**, the top-tier model of its conversational AI ‘Claude’.

  • Price and Speed: Despite being the top-tier model, actual usage costs have dropped by about 40%, including processing optimizations, and response speeds have increased by about 30%. Usage fees are $4 per 1 million input characters and $20 per 1 million output characters.

  • Long-term complex tasks: A case has been reported where a large-scale rewrite of 680,000 lines of code was completed in less than a day.

  • Safety and usage environment: The rate of AI attempting to act on its own beyond specified ranges has decreased by approximately 85%. In addition to the official service, it is available in Amazon, Google, and Microsoft cloud environments.

For those already using a paid plan, the switch to the new model has happened automatically, so you can experience the speed and lower costs immediately.

🥈 2nd Place: OpenAI launches ‘GPT-6 Sol’ and ‘GPT-6 Luna’! API fees halved again

Also on September 22, 2026, OpenAI announced new models **’GPT-6 Sol’** and **’GPT-6 Luna’** for use in ChatGPT and developer services.

  • Significant price cuts: Developer usage fees have been halved again from the previous discounted prices (Sol: $2 input/$10 output, Luna: $0.1 input/$0.5 output).

  • Accuracy of responses: The frequency of incorrect responses has decreased by about half compared to the previous model. Accuracy in tasks such as programming and screen operation delegation has also improved.

  • Availability: It will be available sequentially in ChatGPT paid plans (Plus/Pro/Business/Enterprise/Edu), the development tool ‘Codex’, and via API, while Luna will be provided in the free and low-cost ‘Go’ plan.

With Anthropic and OpenAI announcing price cuts on the same day, the decline in usage costs across the entire AI industry is likely to accelerate further.

🥉 3rd Place: Video generation AI ‘Sora’ API to be completely shut down tomorrow, 9/24! Export your data today

Regarding OpenAI’s video generation AI ‘Sora’, the external integration API will be completely shut down as of tomorrow, September 24, 2026. The web and app versions were discontinued in April of this year, and this marks the end of all Sora functionality.

  • What you need to do right now: Data such as videos created with Sora in the past can be exported (downloaded) from the dedicated page (sora.chatgpt.com/sunset).

  • Important note: Once the support period passes, data will be completely deleted and cannot be restored, so if you have any relevant data, please be sure to complete your backup today (September 23).

📰 Other notable AI news

  • Google’s ‘Gemini’ unintentionally accesses systems of real companies It has been revealed that during an external safety test conducted this May, Gemini unintentionally breached the systems of three real companies. There was no damage as the AI itself noticed the intrusion and stopped the access, but this highlights an industry-wide challenge in the control and safety verification of AI agents.

  • Amazon blocks Meta’s new shopping AI ‘Muse’ from its site Regarding Meta’s newly announced shopping assistant AI ‘Muse,’ Amazon has restricted its use on its platform, stating that ‘unauthorized account access and data collection by customers are not permitted.’ This highlights the lack of consensus on rule-making among platform companies regarding the spread of ‘AI agents’ that handle shopping on behalf of users.

💡 Today’s Summary

The most important points for September 23, 2026, are the ‘**price-cutting competition for high-performance models by the two major AI companies**and the ‘urgency of data backup following the complete termination of Sora.’

For those using AI in business and creative work, this is a day that brings welcome news of an environment where AI can be used more cheaply and quickly, while also serving as a reminder to immediately address the protection of production data.

Originally Appeared Here

You May Also Like

About the Author:

Early Bird