SpaceXAI Grok 4.5 undercuts Opus on price in AI cost war

The News

SpaceXAI has released Grok 4.5, the newest version of the model built by Elon Musk's company, pitching it as a cheaper rival to the most capable systems from Anthropic and OpenAI. The launch landed on Wednesday, 8 July, just weeks after the firm went public.

Musk called it "an Opus-class model, but faster, more token-efficient and lower cost", adding that Grok 4.5 is "roughly comparable to Opus 4.7, but much faster". The company claims the model delivers twice the token efficiency of other leading systems.

Pricing is where the pitch sharpens. Grok 4.5 costs $2 per million input tokens and $6 per million output tokens. That undercuts Anthropic's Opus 4.7, priced at $5 per million input tokens and $25 per million output tokens. It also sits below OpenAI's premium Sol model at $5 input and $30 output, though it lines up closely with OpenAI's cheaper Luna tier at $1 input and $6 output.

Why It Matters

The launch confirms that the frontier race has shifted from raw capability to cost per unit of work. A year ago the contest was which lab could top a benchmark. Now the pitch is who can match that quality at a fraction of the price, and the timing is pointed: OpenAI is reportedly lining up its own GPT 5.6 release the very next day.

Price compression at the top end tends to cascade. When GPT-4 launched in March 2023, running a serious application on a frontier model was a luxury reserved for well-funded teams. Each later round of cuts pulled a fresh tier of builders into the market. Grok 4.5's output pricing at roughly a quarter of Opus 4.7's rate is the sort of move that forces every rival to answer, and fast.

There is a structural signal too. SpaceXAI is a newly listed company now competing head-on with Anthropic and OpenAI, both still privately held. A public model maker faces quarterly scrutiny that private labs do not, which makes such aggressive pricing a doubly bold wager.

Indian Angle

For India's developer economy, token pricing is not a footnote, it is the business case. Halving output costs while claiming Opus-class quality changes the arithmetic for the thousands of Indian SaaS firms, fintech teams and services companies wiring language models into products billed in rupees but paid for in dollars. At $6 per million output tokens, a workload that cost a young company in Bengaluru or Pune a punishing monthly bill on premium models suddenly looks viable at scale.

It also raises the pressure on India's own model builders. Sarvam AI, backed under the IndiaAI Mission, and Krutrim, the Ola-founded venture, have pitched sovereignty and India-specific language coverage as their edge. Global price cuts erode the cost argument for picking a domestic model, pushing local players to compete harder on multilingual quality and data-residency guarantees rather than sticker price alone.

Regulators are watching the dependency question. MeitY has been nudging enterprises and public-sector bodies toward local compute and hosting. Cheaper foreign frontier models make that policy harder to sell, because procurement teams at Indian banks and insurers will struggle to ignore the savings when a listed foreign supplier quotes rupee-friendly rates.

FAQ

When was Grok 4.5 released?

SpaceXAI released Grok 4.5 on Wednesday, 8 July 2026, a few weeks after the company listed publicly. Elon Musk announced the model the same day, framing it as a faster, cheaper alternative to Anthropic's Opus 4.7.

How does the pricing compare?

Grok 4.5 costs $2 per million input tokens and $6 per million output tokens. Opus 4.7 is priced at $5 input and $25 output, while OpenAI's Sol sits at $5 input and $30 output. OpenAI's cheaper Luna tier matches Grok closely at $1 input and $6 output.

What does "Opus-class" mean?

It is Musk's shorthand for capability parity with Anthropic's top Opus model. He said Grok 4.5 is "roughly comparable to Opus 4.7, but much faster", with the company claiming twice the token efficiency of rival systems.

What does it mean for Indian startups?

Lower output pricing improves the unit economics of AI features for Indian SaaS and fintech firms billing in rupees, while raising competitive pressure on domestic model builders such as Sarvam and Krutrim.

Where can I read the original announcement?

TechCrunch reported the launch. The source link sits in the attribution paragraph below.

This story was reported by TechCrunch. Read the full original coverage at TechCrunch.