SpaceXAI just shipped Grok 4.5, its first model release since the company went public a few weeks ago. According to TechCrunch AI, the company pitched the model in a Wednesday blog post as a workhorse built to handle the exact tasks the industry keeps trying to automate: coding, app-building, office and clerical work, research, and writing. Founder Elon Musk went further on X, calling it an “Opus-class model, but faster, more token-efficient and lower cost.”
That’s the headline claim, and it’s a bold one. Opus is Anthropic’s model built for heavy, complex reasoning. Musk is saying Grok 4.5 plays in that league while costing a fraction of the price.
What SpaceXAI is claiming
The pitch comes down to three numbers and one comparison:
- Token efficiency: SpaceXAI says Grok 4.5 has “twice greater token efficiency” than other leading models.
- Pricing: $2 per million input tokens and $6 per million output tokens.
- The benchmark: Musk’s internal read is that Grok 4.5 is “roughly comparable to Opus 4.7, but much faster.”
Musk also noted the release follows “strong positive feedback from customers” in a beta program, with public availability rolling out the day after the announcement.
Why the price tag matters
Token cost has become a real pain point for anyone running AI at scale, and this is where Grok’s numbers get interesting. Look at how SpaceXAI’s pricing stacks up against the competition TechCrunch AI cited:
- Grok 4.5: $2 input / $6 output per million tokens
- Opus 4.7: $5 input / $25 output per million tokens
- OpenAI Sol (top tier): $5 input / $30 output per million tokens
- OpenAI Luna (cheapest): $1 input / $6 output per million tokens
If Grok 4.5 genuinely performs near Opus 4.7 at roughly a quarter of the output cost, that’s a serious value argument. Output tokens are where bills balloon on agentic and long-context workloads, so a jump from $25 to $6 changes the math for a lot of teams.
What stands out here is the positioning. SpaceXAI isn’t claiming best-in-class. Its own benchmark charts, per TechCrunch AI, show Grok competitive with top models but just short of the very top. The play is capability-per-dollar, not raw supremacy.
The catch
Benchmarks and blog posts are marketing until independent testing catches up. “Twice greater token efficiency” and “roughly comparable to Opus 4.7” are SpaceXAI’s own assessments. The real test is whether that efficiency survives contact with messy production workloads, not curated evals. Efficiency claims have a habit of shrinking once you point them at real code and real documents.
Still, the pricing alone forces a response. When a credible model lands at these rates, it puts pressure on every competitor’s cost structure.
The bigger picture
This drops into a packed release week. OpenAI is set to ship GPT 5.6 on Thursday, which it’s calling its “strongest model yet.” TechCrunch AI reports that launch had previously been held back by the Trump administration over security concerns before getting cleared.
So within roughly 24 hours, the market gets a cheaper Opus-class contender from SpaceXAI and OpenAI’s most powerful model to date. The competitive gap at the frontier keeps compressing, and the fight is shifting from “who’s smartest” toward “who’s smart enough at the lowest cost.”
What to watch
A few things worth tracking over the next week:
- Independent benchmarks. Wait for third-party evals before treating the Opus-class claim as settled.
- Real-world efficiency. Does the 2x token-efficiency figure hold on your own workloads, or just on SpaceXAI’s tests?
- Competitor pricing moves. Watch whether Anthropic and OpenAI adjust rates in response.
If you’re picking models on a budget, Grok 4.5 is worth a hands-on trial the moment it’s live. Just verify the claims against your own tasks before you migrate anything. Full details are in the original TechCrunch AI report.