SpaceXAI Grok 4.5 is 11th on Openrouter Token Share

Grok 4.5 (1.5 trillion parameter) at 11 will likely crack the top ten LLM on open router. Based on the number of models they are releasing. They could have new models in the 50% frontier coding score but at prices lower than Anthropic for Opus 5. Later versions of Grok 5.X could push out Opus 5. The battle for the workhorse value model would continue into Q4 2026 and Q1 2027.

Screenshot

SpaceXAI will be releasing a 2 trillion parameter Grok 4.6 will be released in 2 weeks. It should benchmark at about the Kimi K3 level but with more efficiency. Kimi K3 is third on the LLM usage ranking.

Grok 4.7 will be in 4 weeks and will again have more parameters.

Grok 5 might be in 8 weeks.

SpaceXAI will then have multiple models in top ten usage. They could be pushing Google Gemini as the clear third place and being a relevant third place. Currently, third place is not relevant. Anthropic is first with $70-80 billion ARR and OpenAI second with $40 billion ARR. In the best case they could be a solid third and perhaps pushing Openai for second for AI spend. However, SpaceX-Cursor would slow OpenAI spending growth first and then half the price get to $20 billion per year ARR up from $4-5 billion ARR now.

With Grok 5 and Grok 5.1 they could choose to opensource Grok 4.5 and Grok 4.6 (kimi k3) level. Those could become the american open source model and the grok 5, grok 5.1 would be fighting at the opus 5 but maybe half the price.

They would critically need composer 3 and other cursor harness improvements and helping developers get 95% of the work done at lowest cost

Similar Posts

Leave a Reply

Your email address will not be published. Required fields are marked *