GROK
Grok 4.5 promises Opus-level power at a bargain.
Grok is coming for Claude's coding crown.
SpaceXAI just debuted Grok 4.5. This is the first model since the company’s IPO and its acquisition of the AI agent, Cursor. Elon Musk is touting this uplevel as a faster, cheaper option for coding and in-depth knowledge work.
AKA: this ain’t your average chatbot.
What’s interesting here is that Grok was trained on the same compute capacity that xAI leases out to Anthropic and Google. So while the company has diversified income streams by renting its technology to competitors, SpaceXAI hasn’t dropped out of the race. It's just playing both sides.
Musk refers to Grok 4.5 as an “Opus-class” model that can handle the same degree of agentic work as Claude. But it’s much more affordable for the end user.
Crunching numbers:
Grok 4.5 costs $2 per million input tokens and $6 per million output tokens.
Anthropic's Opus 4.8 is significantly pricier, at $5/million for input and $25/million for output.
OpenAI works with tiered rates. Sol sits at $5/million input tokens and $30/million output tokens. But Luna, its least expensive model, is only $1/million input and $6/million output.
Model Rivalry
The ink hadn’t dried on SpaceXAI’s latest announcement when a second launch stole the headlines.
Less than 24 hours after Grok 4.5 came on the scene, GPT 5.6 was released. Dubbed OpenAI’s “strongest model yet,” the team had to hold off on rolling out their latest version after the US government voiced safety concerns.
Then, conveniently, they were able to release GPT 5.6 immediately after Grok hit the market.
Which model is the better deal is still TBD. But no matter how you spin it, the math is clear.
Two rivals + one week = zero chill.
See ya when the next model drops (probably tomorrow). - TL