xAI, Cursor Ship Grok 4.5 at 24% of Opus's Output Price, Aimed Only at Coders, Lawyers, Bankers
TL;DR
xAI and Cursor ship Grok 4.5 for coding, legal and finance, priced at $6 per million output tokens — 24% of Opus 4.8's rate.
xAI (operating as SpaceXAI after being folded into Elon Musk's SpaceX) unveiled Grok 4.5 on July 8 and opened it to the public on July 9. It is the first model co-trained with Cursor since SpaceX's $60 billion acquisition of the coding IDE closed a few weeks earlier. Pricing: $2 per million input tokens, $6 per million output tokens. Anthropic's Claude Opus 4.8 charges $5 and $25 for the same. Grok 4.5's output rate is 24% of Opus.
The model runs on xAI's new 1.5-trillion-parameter V9 base, roughly three times the size of the v8-small architecture behind Grok 4.3. Cursor folded its IDE coding telemetry into the supplemental training stage. xAI positioned it not as a general chatbot but for software engineering, legal work, financial analysis and AI agents, and pitched long-horizon agentic runs as the main upgrade. Musk wrote on X: "It is an Opus-class model, but faster, more token-efficient and lower cost." xAI said Grok 4.5 ranked #1 on Harvey's Legal Agent Benchmark and beat Claude Opus 4.8 on several coding benchmarks.
The twist is in the distribution channel. Before the acquisition, Cursor was one of Anthropic's largest third-party resellers — a large share of Claude Sonnet and Opus IDE calls came through Cursor's subscription pool. Cursor subscribers are now defaulted to Grok 4.5, and Anthropic loses its single biggest retail pipeline for its most expensive tier overnight. Bloomberg reported Grok 4.5 shipped simultaneously in Grok Build, every Cursor plan, and the SpaceXAI console, with EU rollout held back. A premium latency-tuned tier at $4 / $18 targets quant funds and investment-bank document review.
If it works, xAI uses Cursor's distribution and SpaceX's balance sheet to push Anthropic out of the coding-IDE channel entirely. If it doesn't, $60 billion bought one model that undercuts Opus by 60% in three verticals.
via Bloomberg / The Next Web / The Tech Portal
The model runs on xAI's new 1.5-trillion-parameter V9 base, roughly three times the size of the v8-small architecture behind Grok 4.3. Cursor folded its IDE coding telemetry into the supplemental training stage. xAI positioned it not as a general chatbot but for software engineering, legal work, financial analysis and AI agents, and pitched long-horizon agentic runs as the main upgrade. Musk wrote on X: "It is an Opus-class model, but faster, more token-efficient and lower cost." xAI said Grok 4.5 ranked #1 on Harvey's Legal Agent Benchmark and beat Claude Opus 4.8 on several coding benchmarks.
The twist is in the distribution channel. Before the acquisition, Cursor was one of Anthropic's largest third-party resellers — a large share of Claude Sonnet and Opus IDE calls came through Cursor's subscription pool. Cursor subscribers are now defaulted to Grok 4.5, and Anthropic loses its single biggest retail pipeline for its most expensive tier overnight. Bloomberg reported Grok 4.5 shipped simultaneously in Grok Build, every Cursor plan, and the SpaceXAI console, with EU rollout held back. A premium latency-tuned tier at $4 / $18 targets quant funds and investment-bank document review.
If it works, xAI uses Cursor's distribution and SpaceX's balance sheet to push Anthropic out of the coding-IDE channel entirely. If it doesn't, $60 billion bought one model that undercuts Opus by 60% in three verticals.
via Bloomberg / The Next Web / The Tech Portal
