Story perspectives

AI Arms Race Escalates: New Models Boost Efficiency, Costs Surge

7/13/2026

Summary

Story summary

OpenAI launched GPT-5.6 on Thursday, touting 54% higher token efficiency for agentic coding.
SpaceXAI introduced Grok 4.5, claiming twice the token efficiency and lower cost than rivals.
Meta Platforms released Muse Spark 1.1 with attractive pricing while criticizing competitors’ margins.
OpenRouter raised more than $100 million in May to expand its model-routing services.
Anthropic’s Opus and Fable remain costly, leading firms to set tighter AI spending limits after monthly invoices hit millions of dollars.

1 of 1

Story summary

OpenAI launched GPT-5.6 on Thursday, touting 54% higher token efficiency for agentic coding.
SpaceXAI introduced Grok 4.5, claiming twice the token efficiency and lower cost than rivals.
Meta Platforms released Muse Spark 1.1 with attractive pricing while criticizing competitors’ margins.
OpenRouter raised more than $100 million in May to expand its model-routing services.
Anthropic’s Opus and Fable remain costly, leading firms to set tighter AI spending limits after monthly invoices hit millions of dollars.