xAI released Grok 4.5 on Wednesday, pricing the model at $2 per million input tokens and $6 per million output tokens — a rate that undercuts Anthropic's Claude Opus 4.7 by more than half on input and roughly 4x on output. Elon Musk described the release as an Opus-class model, but faster and cheaper, positioning Grok directly against the frontier tier rather than against mid-market offerings. It is the first model xAI has shipped since the company went public several weeks ago.
The pricing gap is the story. Opus 4.7 costs $5 per million input tokens and $25 per million output tokens. OpenAI's top-tier Sol runs $5 input and $30 output; its cheaper Luna sits at $1 input and $6 output. Grok 4.5 lands between those two OpenAI tiers on input but matches Luna's output rate — while claiming Opus-class capability, not Luna-class.
xAI also asserts that Grok 4.5 has 'twice greater token efficiency' than other leading models, which, if it holds up in production workloads, compounds the raw price advantage. Token cost has become a live P&L issue for enterprise buyers running agents at scale, and efficiency gains translate directly to margin.
“Based on strong positive feedback from customers in our beta test program, @SpaceXAI will make Grok 4.5 available to the public tomorrow. It is an Opus-class model, but faster, more token-efficient and lower cost”— Elon Musk, xAI founder
Key facts
- 01xAI priced Grok 4.5 at $2 per million input tokens and $6 per million output tokens, less than half the rate of Claude Opus 4.7.
- 02Anthropic's Opus 4.7 runs $5 per million input and $25 per million output — roughly 4x Grok 4.5's output cost.
- 03xAI claims Grok 4.5 has 'twice greater token efficiency' than other leading models.
- 04OpenAI is set to release GPT 5.6 on Thursday, after Trump administration security review delayed the launch.
- 05Grok 4.5 is xAI's first release since the company went public several weeks ago.
Musk posted the announcement on X, the xAI subsidiary. Benchmark charts released alongside the blog post showed Grok tracking near the top of the field on standard evaluations, though not consistently best-in-class. The company's pitch is not that Grok wins every benchmark — it is that Grok wins on price-per-capability.
Musk followed with a direct comparison to Anthropic's flagship.
“Our internal assessment is that Grok 4.5 is roughly comparable to Opus 4.7, but much faster. The combination of capability, faster speed and lower cost is what makes it competitive.”— Elon Musk, xAI founder
Anthropic has not responded to the pricing challenge. Opus 4.7 remains Anthropic's premium tier and is priced accordingly, with the company betting that reliability on long-horizon agentic tasks justifies the premium. Whether Grok 4.5 matches that reliability outside of curated benchmarks is the open question — internal assessments from a model vendor tend to flatter the model.
The release also lands one day before OpenAI is scheduled to ship GPT 5.6 on Thursday, which the company has called its 'strongest model yet.' That launch was previously delayed by the Trump administration over security review concerns. Two frontier releases in two days puts pricing and capability disclosures from all three top labs on the table in the same news cycle.
For enterprise buyers, the math is straightforward. A workload burning 100 million output tokens a month costs $2,500 on Grok 4.5 and $2,500 on OpenAI Luna, versus $3,000 on OpenAI Sol and $2,500 on Anthropic Opus 4.7 — wait, Opus 4.7 runs that same workload at $2,500 on output alone at a much higher per-token rate, meaning at 100M tokens Grok costs $600 and Opus costs $2,500. The gap widens sharply at scale.
Skeptics will note that xAI's benchmark disclosures are self-published and that 'Opus-class' is a marketing claim until third-party evaluators confirm it. Grok's prior releases have shown strong scores on specific benchmarks but uneven performance on agentic tasks where Anthropic's models have set the bar. The efficiency claim, in particular, needs independent verification before buyers redesign their token budgets around it.
If Grok 4.5 holds up under third-party testing, xAI has done something the frontier labs have avoided: broken the price ceiling on top-tier models. Anthropic and OpenAI have both trended toward higher prices with each capability jump, betting that enterprise willingness-to-pay scales with intelligence. xAI is testing the opposite hypothesis — that price compression at the top tier forces the incumbents to either match or cede volume. The next several quarters of API revenue disclosures will show which bet was right.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
