After SpaceXAI released Grok 4.5, Musk described it on X as an "Opus-level product, faster and cheaper," and later added that "it seems to rank first in at least some benchmark tests"; however, Musk later admitted that this comparison was against the previous generation Claude Opus 4.8, not the current one.
Grok 4.5 costs $2 for input, while Claude Opus 4.8 costs $5
According to SpaceXAI's official pricing and Artificial Analysis data, Grok 4.5's API pricing is $2 per million tokens for input and $6 for output; Claude Opus 4.8 is priced at $5 for input and $25 for output.
In terms of coding task costs, Artificial Analysis estimates Grok 4.5 at $2.49 per task (using about 1.9 million tokens), while Claude Code's Fable 5 costs $11.80 per task (using about 7.2 million tokens).
Musk's core competitive advantage for Grok 4.5 is positioning it as a product where the engineering team increasingly emphasizes cost-effectiveness in task completion rather than solely pursuing the highest intelligence level.
Artificial Analysis Intelligence Index: Grok 4.5 ranks fourth
(Source: Thibault Jaigu)
Based on independent testing data from Artificial Analysis and Snorkel AI, here are Grok 4.5's performance results:
Artificial Analysis Intelligence Index: Grok 4.5 scores 54, ranking fourth (behind Fable 5, GPT-5.5, and Claude Opus 4.8).
Coding Agent Index (Grok Build test): Score of 76, comparable to GPT-5.5 in Codex, slightly below Fable 5's performance in Claude Code.
Snorkel AI GDPval+ benchmark (about 2,000 professional task samples): Grok 4.5 has a pass rate of 29%, leading GPT 5.5 (22%) and Opus 4.8 (21%), with notable advantages in legal, educational, and healthcare tasks. Note: During testing, Grok 4.5 was used in conjunction with SpaceXAI's self-developed Grok Build agent.
Cursor agrees to acquire SpaceXAI for $60 billion in all-stock deal, expected to close in Q3 2026
According to reports, AI startup Cursor has agreed to acquire SpaceXAI in a $60 billion all-stock transaction, with completion expected in the third quarter of 2026.
Cursor states that the collaboration began during the training phase, utilizing real data from engineers writing, reviewing, and debugging code; the training used tens of thousands of NVIDIA GB300 GPUs.
Grok 4.5 is now available through SpaceXAI's main server, Grok Build agent, and Cursor in all packages, with EU region deployment expected in mid-July.
Frequently Asked Questions
How does the pricing of Grok 4.5 compare to Claude Opus 4.8?
According to SpaceXAI's official pricing, Grok 4.5 costs $2 per million tokens for input and $6 for output; Claude Opus 4.8 costs $5 for input and $25 for output; each coding task costs $2.49 (Fable 5 in Claude Code costs $11.80).
How do third-party testing agencies evaluate Grok 4.5's ranking?
Based on the Artificial Analysis Intelligence Index, Grok 4.5 ranks fourth, behind Fable 5, GPT-5.5, and Claude Opus 4.8; in Snorkel AI's GDPval+ benchmark, Grok 4.5 has a pass rate of 29%, leading GPT 5.5 (22%) and Opus 4.8 (21%).
Is there benchmark test support for Musk's claim that Grok 4.5 is close to Opus?
According to Cryptopolitan reports, Musk's statement that Grok 4.5 is "close to or even surpassing Opus" is not based on published benchmark tests but rather on the company's evaluators' impressions; Musk later clarified that the comparison was against the previous generation Claude Opus 4.8, not the current one.
