Grok 4.6 Just Matched GPT-5.6 Sol at a Fraction of the Price

Grok 4.6 is now available after SpaceXAI introduced its latest frontier AI model, targeting long-running tasks, coding, technical reasoning, and engineering work.

The company says Grok 4.6 combines stronger intelligence with fast inference and competitive pricing, while improving its ability to complete difficult tasks on the first attempt.

Grok 4.6 targets agentic and coding workloads

SpaceXAI trained Grok 4.6 across a broad range of agentic reinforcement learning tasks, including general coding, knowledge work, and specialized technical environments.

Compared with Grok 4.5, the new model should perform better on longer-running workflows and produce stronger initial responses when handling complex tasks.

Grok 4.6 matches GPT-5.6 Sol in intelligence testing

Grok 4.6 matched GPT-5.6 Sol on the Artificial Analysis Intelligence Index, placing behind Claude Opus 5 and Fable 5 Max.

The index combines results from nine AI benchmarks.

On GDPval-AA v2, Grok 4.6 reached an Elo score of 1,753, placing it behind only Claude Opus 5. On AA-Briefcase, a benchmark focused on long-horizon agentic knowledge work, the model scored 1,577 Elo and again ranked behind Claude Opus 5.

SpaceXAI prices Grok 4.6 from $2 per million tokens

The standard Grok 4.6 model runs at roughly 80 tokens per second and costs $2 per million input tokens and $6 per million output tokens.

SpaceXAI also offers a faster version priced at $4 per million input tokens and $12 per million output tokens.

The company is positioning Grok 4.6 as a lower-cost option for developers who need frontier-level reasoning and agentic capabilities.

Grok 4.6 arrives in Cursor, Grok Build, and the API

Grok 4.6 is available through Cursor, Grok Build, and the SpaceXAI API.

SpaceXAI is also doubling the included Grok 4.6 usage for Cursor and Grok Build users during the model’s first week.

In other AI news, Microsoft recently released MAI-Code-1.1-Flash, while OpenAI launched GPT-5.6-Cyber.