Musk "copied" Zhipu

```

xAI releases Grok 4.5, challenging Anthropic with "Opus-level" capabilities, and targeting Chinese open-source model strategies with high cost-effectiveness.

On Wednesday, Elon Musk’s xAI released its latest model Grok 4.5, claiming to achieve intelligence at the same level as Anthropic’s flagship model for less than one fifth of its output token price.

This pricing strategy closely resembles the previous approach by Zhipu’s GLM-5.2, which matched closed-source leading models with 82% lower pricing, prompting market observers to remark that leading Silicon Valley labs are starting to adopt the cost-effectiveness tactics used by Chinese open-source players. Musk himself posted on X:

Grok 4.5 is an Opus-level model, but faster, more efficient with tokens, and cheaper.

He further stated that internal evaluations show Grok 4.5’s overall capabilities are roughly on par with Opus 4.7, but it is "much faster." The combination of capability, speed, and low cost forms the core competitive logic of this model.

According to xAI’s announced pricing, Grok 4.5 is priced at $2 per million input tokens and $6 per million output tokens. By contrast, Anthropic Opus 4.7’s input and output prices are $5 and $25 per million tokens, respectively—a huge gap.

If this pricing strategy delivers on its claimed capabilities, it will directly impact procurement decisions for AI enterprise customers. Meanwhile, OpenAI plans to release its latest flagship model GPT 5.6 on Thursday, making this week a dense window period for AI model launches.

Cost-effectiveness route: token efficiency becomes a core selling point

xAI disclosed in its blog that Grok 4.5’s token efficiency is twice that of other leading models.

If this metric holds in practical applications, enterprise-level inference costs will fall directly, which is highly attractive to enterprise users for whom token spending is increasingly a core concern in AI procurement.

From a horizontal comparison, OpenAI’s tiered pricing system charges $5 and $30 per million input/output tokens respectively for the high-end Sol version, and $1 and $6 for the low-end Luna version.

Grok 4.5’s pricing is significantly lower than major competitors, and in Musk’s narrative framework, it corresponds to model capabilities close to Anthropic’s flagship level.

"White-haired stock god" Serenity commented on X that what xAI is doing is somewhat similar to what Chinese AI vendors have always done: competing via price and adoption rates.

Just a few weeks ago, Zhipu’s open-source model GLM-5.2, with pricing about 72% to 82% lower, scored 74.4 on the leading programming benchmark FrontierSWE, less than one percentage point behind Anthropic Opus 4.8’s 75.1, while surpassing GPT-5.5’s 72.6.

JPMorgan later summarized this phenomenon in a report: mature AI intelligence continues to compress pricing, but upgrades in frontier capabilities can still justify a premium, and the monetization logic of AI models is exhibiting structural differentiation.

The emergence of Grok 4.5 further complicates this competitive landscape: as leading US labs begin to enter the market with pricing strategies close to open-source models, institutions like Anthropic, which rely on high-priced closed-source models, will feel pressure on both capability and cost dimensions.

Programming agent market: Cursor acquisition logic surfaces

In product positioning, xAI describes Grok 4.5 as a "mainstay model" for daily knowledge work, covering scenarios such as programming and application development, office documentation, research, and content writing. This echoes its previous strategic logic of acquiring the code editing tool Cursor.

Jukan, an analyst at investment research firm Citrini Research, noted in a comment that at the recent ICML conference, he heard a noteworthy Grok multi-pole logic: Claude Code is currently leading the programming agent market, not simply due to model quality, but thanks to an early, large-scale user base. More users bring more real programming data, which in turn improves model quality, forming a flywheel effect.

Jukan believes the core value of xAI’s acquisition of Cursor lies exactly here:

Cursor likely has a larger real user base and code dataset than OpenAI Codex; if xAI can effectively train and utilize this data, surpassing Codex may only be a matter of time.

Risk Warning and DisclaimerThe market entails risk, investment should be cautious. This article does not constitute personal investment advice, nor does it take into account the specific investment objectives, financial situation, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article are suitable for their own situation. Invest at your own risk. ```