- Google has released the new model Gemini 4 Argon, and the company says it matches or beats the top models from OpenAI and Anthropic on several benchmarks. A benchmark is a standard that compares AI performance using the same tests.
- Google priced Argon per token, and according to the source, it is up to 80 percent cheaper than competitors during its introductory period. A token is the unit AI uses to charge for processing text. The source also says it remains cheaper than competitors after the introductory period ends.
- Meta has released Muse Wonder Full, and in Meta's own tests it says it can compete with GPT-6 Astra, Claude Opus 5.5, and Claude Fable 5.1 on several benchmarks. Results from independent organizations are not given in the source.
- According to the source, Meta's Muse agent assistant gained popularity in a short time, but concerns about security and privacy are also growing.
O: Useful for AI service developers who need to compare token prices of Google Gemini 4 Argon and rebuild a monthly budget by placing the introductory-period discount prices and the regular prices side by side.
X: The discount of up to 80 percent applies only during the introductory period, and the source does not give a specific regular price, so the discount rate should not be applied directly to long-term costs.