SpaceXAI Grok 4.5 Launches With A Lower Token Price

SpaceXAI has released Grok 4.5 with a pricing claim aimed at heavy AI users: the company said the model costs $2 per 1 million input tokens and $6 per 1 million output tokens.

The SpaceXAI Grok 4.5 pricing claim puts token efficiency at the centre of the launch, but the company did not publish independent benchmark validation or named customer adoption data.

The company described Grok 4.5 as a workhorse model for coding, email writing, presentations, office tasks, research and other knowledge-based work.

SpaceXAI said the model can handle those tasks with about half the cost of peers because it has twice greater token efficiency than other frontier model labs.

SpaceXAI also released benchmark claims with the launch.

The published account described the results as competitive with major rival models while still falling short of their top performance, so the company-owned efficiency claim carries more weight than a pure capability lead.

Musk Said Grok 4.5 Will Be Public After Beta Feedback

Elon Musk said on X on July 8, 2026, that SpaceXAI would make Grok 4.5 available to the public the next day after positive feedback from beta customers.

Musk described the model as Opus-class, faster, more token-efficient and lower cost.

Musk also said SpaceXAI's internal assessments showed Grok 4.5 was roughly comparable with Opus 4.7 while generating results faster.

That statement came from company-controlled testing, and SpaceXAI did not include a third-party lab result, full benchmark methodology or named enterprise customer reference.

The launch arrives during a busy week for AI model releases.

OpenAI announced GPT-5.6 Sol on the same day and also launched GPT-Live, a family of models optimised to process spoken instructions.

Frontier Model Prices Give Grok 4.5 Its Main Comparison

SpaceXAI listed Grok 4.5 at $2 for 1 million input tokens and $6 for 1 million output tokens.

According to SiliconANGLE, the pricing comparison set Opus 4.7 and 4.8 at $5 for 1 million input tokens and $25 for 1 million output tokens, Anthropic's Fable 5 at $10 for 1 million input tokens and $50 for 1 million output tokens, OpenAI's GPT-5.6 Sol at $5 for 1 million input tokens and $30 for 1 million output tokens, and the lower-cost Luna model at $1 for 1 million input tokens and $6 for 1 million output tokens.

Those figures make Grok 4.5 cheaper than several named frontier models in the published pricing comparison, but not cheaper than every OpenAI tier.

The comparison also separates price from capability: SpaceXAI is presenting lower token cost and speed as the competitive route rather than claiming a clear benchmark lead across the market.

SpaceXAI Does Not Include Third-Party Benchmark Proof

The release gives SpaceXAI a fresh model for coding, office work, research and voice-adjacent AI competition, while OpenAI and Anthropic remain the named comparison set.

Enterprise buyers still need workload-specific evidence before treating lower token prices as lower operating cost.

SpaceXAI did not disclose independent benchmark validation, full test methodology, named enterprise customers, public beta customer totals, service-level terms, data-handling commitments or deployment restrictions for Grok 4.5.