Grok 4.6
Points and comments are a snapshot, not live.
Grok 4.6 matches GPT-5.6 Sol on composite AI benchmark, available in Cursor and Build.
xAI releases Grok 4.6, focusing on long-running agents for coding and knowledge work. It scores 61 on the Artificial Analysis Intelligence Index, tied with GPT-5.6 Sol and one point behind Fable 5 Max. Benchmarks include DeepSWE v1.1 (65.9%), CursorBench v3.2 (69.9%), and FrontierCode v1.1 (61.3%). Training used curated model-generated data and RL on agentic tasks. Pricing starts at $2 per million input tokens and $6 per million output tokens. Available in Cursor, Grok Build, and API with doubled usage for the first week.
What commenters are saying
Commenters are impressed by Grok 4.6's benchmark parity with Fable 5 and GPT-5.6 Sol at lower cost. Some question the timing of releases after Fable, suggesting synthetic RL tasks and scale as the common technique. A split: those who find Grok 4.5 between Sonnet and Opus level, and those who find it Opus-level for coding. Several users report switching to Grok for cost savings. One commenter notes the Cursor acquisition hasn't fully closed, creating confusion about the partnership.