SpaceXAI Launches Grok 4.7: A Bigger Foundation Model Priced at the Same $2/$6 as Grok 4.6

SpaceXAI has released Grok 4.7, its newest flagship model aimed at coding, agentic workflows, and knowledge work. The model is built on a larger base and trained with a longer reinforcement learning run, yet it ships at the identical price and speed of Grok 4.6. It is available immediately as a hosted model through the xAI API, Cursor, Grok Build, OpenRouter, Vercel, and Cloudflare, making it one of the most accessible premium coding models on the market today.

What Changed Under the Hood

SpaceXAI highlights four upgrades over Grok 4.6. First, the model runs on a new, larger base rather than reusing its predecessor's foundation. Second, the reinforcement learning run was extended and weighted toward tasks that take many hours to complete. Third, the company claims stronger self-verification and long-context handling, meaning the model checks its own output more rigorously. Fourth, Grok 4.7 was trained natively to work with the Grok Bot harness for conversational and knowledge work.

On the technical side, the API exposes a 500,000-token context window, a knowledge cutoff of May 2026, and text-and-image input with text output. Reasoning effort can be set to low, medium, high (the default), or xhigh, and the model supports function calling, web search, X search, and code execution through the Responses API and Chat Completions.

Benchmarks: Strong Gains, but Not a Clean Sweep

In the launch comparison against Grok 4.6 High, GPT-5.6 Sol Max, and Fable 5.1 Max, Grok 4.7 improves on Grok 4.6 in every row. The biggest jump comes on Terminal-Bench 4.0, which rose from 20.3% to 38.0%. EEBench climbed 11 points to 64.0%, the best score in the table, and the Harvey Legal Agent Benchmark result of 19.6% far outpaces Fable 5.1 Max's 6.7%.

Still, Grok 4.7 does not lead everywhere. Fable 5.1 Max tops 4 of 7 benchmarks, including Terminal-Bench 4.0 at 57.9%, while GPT-5.6 Sol Max holds the best DeepSWE v1.1 result at 72.7%. Price is where SpaceXAI argues its case: Fable 5.1 Max costs 5x more on input and roughly 8.3x more on output, while GPT-5.6 Sol Max runs 2x more on input and about 3.3x more on output. On a CursorBench 4.0 cost-per-task chart, SpaceXAI places Grok 4.7 at the price-performance frontier. On GDPval, which measures professional knowledge work, Grok 4.7 xhigh scored 1,695 Elo, up from 1,605 for Grok 4.6 high, though Fable 5.1 Max leads at 1,735 and GPT-6 Astra Max scored 1,542.

Safety and Availability

Grok 4.7 ships with a completely new safeguard stack, which SpaceXAI describes as its strongest model yet on refusals and jailbreak resistance. It topped LatchBio's biosafety benchmark at 62.4%. On HackerBench v0.3, SpaceXAI's internal test for risky and malicious cyber tasks, only 3.3% of risky dual-use prompts got through, and the company says the model rarely blocks legitimate security work. Invite-only red-team access is now open to select cybersecurity partners for defense research.

Pricing stays at $2 per million input tokens and $6 per million output tokens. Grok 4.7 is available in Cursor on all plans, is the default model in Grok Build, and is served via the Grok API, OpenRouter, Vercel, and Cloudflare. A Grok 4.7 Fast variant doubles output speed at double the price but runs only in Cursor and Grok Build, not the public API, and is excluded from Grok Build's free tier. A US regional endpoint keeps inference stateside at a 10% premium, and SpaceXAI advises setting a prompt_cache_key for dependable cache hits.

Key takeaways:

• Grok 4.7 retains Grok 4.6 pricing: $2 input and $6 output per million tokens.
• It leads the launch table on EEBench (64.0%) and the Harvey Legal Agent Benchmark (19.6%).
• Fable 5.1 Max still tops 4 of 7 benchmarks, including Terminal-Bench 4.0.
• The API offers a 500,000-token context window and four reasoning levels up to xhigh.
• Just 3.3% of risky dual-use prompts passed HackerBench v0.3.

With rival models holding narrow benchmark leads at far higher prices, the real contest will be whether Grok 4.7's price-performance holds up in production workloads — watch for independent evaluations and partner feedback in the coming weeks.

Meta description: SpaceXAI releases Grok 4.7 with a larger base model, 500K-token context, and unchanged $2/$6 pricing. See benchmarks, safety results, and availability.

Tags: SpaceXAI, Grok 4.7, AI benchmarks, coding models, LLM pricing

Featured image: Abstract technology style — flowing neural network patterns in deep blue and teal gradients, no people or logos.