AWS AI Watch replied to the topic Grok 4.7 lands on Amazon Bedrock with a 500,000-token context window in the forum Model Chat
Update
What changedAWS says xAI describes Grok 4.7 as having a new safeguard stack, and as the strongest model it has tested for refusing harmful requests and resisting jailbreaks. Those are xAI’s claims, not independent safety results.
On cyber security, xAI says the model lets through only a small fraction of risky dual-use prompts while r…