xAI rolls out Grok 4.7 at rock-bottom prices, yet benchmark results show a sizable gap behind Claude and GPT-6
xAI has released Grok 4.7, its most advanced model for coding and knowledge tasks, built on a larger base model with extended reinforcement learning. Priced at $2/$6 per million input/output tokens, it undercuts Western rivals. However, it scores 46 on the Artificial Analysis Intelligence Index versus 53 for Claude Fable 5.1 and GPT-6, and only 26% on Terminal-Bench 4.0, revealing a significant agentic coding gap.
Elon Musk's xAI has rolled out Grok 4.7, the company's most advanced model to date for coding and knowledge-based tasks. The release matters because xAI is positioning the model at aggressive price points while facing stiff competition from rivals that are posting noticeably higher benchmark results.
How Grok 4.7 Was Built
According to xAI, the new model rests on a larger base model and underwent extended reinforcement learning during training. The company also says Grok 4.7 is better engineered to verify its own output, a capability increasingly seen as essential for reliable agentic workflows.
Pricing That Undercuts Western Rivals
Grok 4.7 is priced at $2 per million input tokens and $6 per million output tokens. That puts it much closer to the rates charged by Chinese model providers than to Western frontier labs, and the benchmark results suggest the pricing strategy reflects the model's competitive standing.
On the independent Artificial Analysis Intelligence Index (v4.3.2), which aggregates ten benchmarks, Grok 4.7 scores 46 — a mid-pack result. Claude Fable 5.1 and GPT-6 top the chart with 53 points each. Notably, the model's two highest reasoning levels appear to deliver roughly identical performance, offering little practical benefit for choosing the more expensive tier.
Agentic Coding Is the Weak Spot
The performance gap widens considerably in agentic coding tasks. On Terminal-Bench 4.0, Grok 4.7 manages only 26 percent, while GPT-6 Astra reaches 60 percent and Claude Fable 5.1 hits 55 percent. Even DeepSeek V4.1 Flash, a cheaper alternative, edges ahead at 27 percent.
Developers can access Grok 4.7 through the Grok API, Cursor, and Grok Build.
What to Watch Next
The modest pricing signals that xAI is competing on cost rather than raw capability for now. Whether a future iteration can close the agentic coding gap with Anthropic and OpenAI will likely determine Grok's standing among developers.
Meta description: xAI's Grok 4.7 arrives with $2/$6 per million token pricing, but benchmark scores trail Claude Fable 5.1 and GPT-6 on key coding tests.
Tags: xAI, Grok 4.7, AI benchmarks, agentic coding, LLM pricing
Comments
No comments yet. Be the first to comment.