Top stories
xAI releases Grok 4.5 with aggressive coding-model pricing
- Category: AI
- Status: confirmed
- Sources: xAI announcement, Cursor blog, HN discussion
- Summary: xAI released Grok 4.5 to the public on 2026-07-08, eleven days after a private beta at SpaceX and Tesla. The model is built on V9, xAI's ninth-generation architecture reported at 1.5 trillion parameters, and xAI says it folded real Cursor developer session data (debugging traces, multi-file diffs, user corrections) into training. Reported pricing is $2 per million input tokens and $6 per million output tokens. On the four benchmarks xAI published it beats Opus 4.8 on DeepSWE 1.0 and Terminal-Bench 2.1 and loses on DeepSWE 1.1 (by 6 points) and SWE-Bench Pro (by 4.5 points), and xAI states it uses about 4.2 times fewer tokens than Opus 4.8 on SWE-Bench Pro. It is available now in Grok Build, in Cursor on all plans, and through the SpaceXAI console. Cursor, whose team co-trained the model, says Grok 4.5 and its own Composer 2.5 are different weight classes and both remain available.
- Comments: HN commenters read the model as unusually economical, noting the $2/$6 price against $5/$30 for GPT-5.5 and $5/$25 for Opus 4.8, and flag the benchmark figures as vendor-reported.
- Why it matters: A frontier-class coding model priced well below current OpenAI and Anthropic tiers and available inside Cursor immediately changes the cost calculus for agentic coding workloads.
- Follow-up: Watch independent benchmark reproduction, the token-efficiency claim, and whether the reported Cursor tool-calling gaps get fixed.