SpaceX Grok 4.6 matches OpenAI score at fraction of cost
SpaceX's new Grok 4.6 model matches OpenAI's GPT-5.6 Sol Max with a score of 61 on the Artificial Analysis Intelligence Index. However, Grok 4.6 offers significant cost advantages, charging $2/$6 per million input/output tokens compared to Sol Max's $5/$30. This positions xAI competitively against Anthropic, which leads the index with a 67% probability of having the best model by year-end on Polymarket.

*this image is generated using AI for illustrative purposes only.
SpaceX (NASDAQ: SPCX) released Grok 4.6 on Wednesday, achieving a score of 61 on the Artificial Analysis Intelligence Index. This performance matches OpenAI's GPT-5.6 Sol Max and trails only Anthropic's leading models, according to independent benchmarking firm Artificial Analysis.
The pricing structure for Grok 4.6 is significantly lower than its primary competitor. Grok charges $2 per million input tokens and $6 per million output tokens. In comparison, Sol Max costs $5 for input and $30 for output. This cost differential positions the model to shift the competitive dynamic from pure intelligence benchmarks to price efficiency.
Performance Gains
Grok 4.6 gained five points over Grok 4.5 in little more than a month. It represents a 23-point improvement over Grok 4.3. Artificial Analysis noted that the release returns SpaceX's AI unit to the frontier, with particularly strong performance on agentic work rather than static reasoning.
Anthropic remains the overall leader in the index. Polymarket now assigns xAI a 9% chance of having the best AI model by the end of the year, up from around 3% before the latest release. The market resolves using Arena's leaderboard rather than the Artificial Analysis index. Anthropic holds a 67% probability, while Google has fallen to fourth place behind xAI for the first time this year, with 6%.
Competitive Landscape
Grok's acceleration coincides with challenges faced by rivals. Reuters reported that Alphabet co-founder Sergey Brin has urged key AI staff to focus entirely on Gemini. Gemini briefly reached the frontier last November but subsequently slipped behind. Google delayed its next flagship model by two months following weak internal coding results. Compute shortages, internal disputes, and bureaucracy have also slowed progress.
Investor Implications
Morgan Stanley stated this week that investors may be undervaluing SpaceX's AI business. The bank indicated that further signs of Grok gaining traction through Cursor could help close the valuation gap. Morgan Stanley set its bull case for SpaceX shares at $600. Grok 4.6 is now available through Cursor and SpaceX's API.
What the Numbers Show
The pricing disparity between Grok 4.6 and Sol Max is substantial across both input and output metrics. While both models achieve identical scores of 61 on the Artificial Analysis Intelligence Index, Grok's output token cost ($6) is one-fifth of Sol Max's ($30). This suggests a potential margin advantage for users relying heavily on generative output, where costs typically exceed input expenses in agentic workflows.
How might Grok 4.6's aggressive pricing strategy force competitors like OpenAI and Anthropic to adjust their token costs, potentially triggering a price war in the AI inference market?
To what extent could the integration of Grok 4.6 into Cursor accelerate developer adoption of xAI models compared to traditional API-only distributions?
Will Google's reported internal delays and compute shortages allow xAI to permanently displace Alphabet from the top tier of frontier AI providers by the end of the year?

































