SpaceXAI's Grok 4.5 challenges Opus with 2x token efficiency
Grok 4.5 claims Opus-class performance at a fraction of the cost...
SpaceXAI has launched Grok 4.5, its first major model release since the company went public. In a blog post, SpaceXAI positioned the model as a versatile workhorse capable of coding, app-building, clerical tasks, research, and writing. The company claims that Grok 4.5 offers 'twice greater token efficiency' than leading competitors, potentially addressing the growing concern over inference costs for AI consumers. Pricing is aggressive: $2 per million input tokens and $6 per million output tokens, significantly undercutting Anthropic's Opus 4.7 ($5/$25) and OpenAI's top-tier models (up to $5/$30).
Elon Musk posted on X that Grok 4.5 is an 'Opus-class model' but faster, more token-efficient, and lower cost, based on beta tester feedback. He added that internal assessments show it rivals Opus 4.7 in capability while being much faster. Despite the bold claims, benchmark metrics released by SpaceXAI show Grok 4.5 as competitive but just short of best-in-class performance. The release coincides with a busy week for AI: OpenAI is expected to launch GPT-5.6 on Thursday, which had been delayed by the Trump administration due to security concerns. Grok 4.5 becomes publicly available on Thursday, directly competing with both models.
- Grok 4.5 costs $2/M input and $6/M output tokens, undercutting Opus 4.7 ($5/$25) and OpenAI's Luna ($1/$6).
- SpaceXAI claims 2x token efficiency over leading models, reducing overall inference costs.
- Public release coincides with OpenAI's GPT-5.6 launch on Thursday, creating a direct showdown.
Why It Matters
Affordable Opus-class performance could pressure pricing across the AI industry and democratize access to high-quality models.