Reported SpeechMap scores put Anthropic's 5.5 models at a company high and Grok 4.7 at an xAI low · Digg