Use @grok Build with AI SDK 𝙷𝚊𝚛𝚗𝚎𝚜𝚜𝙰𝚐𝚎𝚗𝚝: 𝚌𝚘𝚗𝚜𝚝 𝚊𝚐𝚎𝚗𝚝 = 𝚗𝚎𝚠 𝙷𝚊𝚛𝚗𝚎𝚜𝚜𝙰𝚐𝚎𝚗𝚝({ 𝚑𝚊𝚛𝚗𝚎𝚜𝚜: 𝚐𝚛𝚘𝚔𝙱𝚞𝚒𝚕𝚍, });

Grok Build just beat Claude Code in a cross-project memory test. The comparison used four small Node.js repositories and three tests, each split across two separate sessions. Results: • Grok Build passed 3/3 tests • Claude Code passed 2/3 tests • Both remembered project-specific commands and coding decisions • Only Grok automatically carried a rule from one repository into another • Grok used 390,848 tokens vs Claude’s 576,863, around 32% fewer • Grok cost $0.41 vs Claude’s $1.05 in this test, around 61% less.

Grok 4.7 just built a GTA-style open-world game from a single prompt in Grok Build You can walk around, drive cars, enter vehicles and explore an entire city It looks insane, and I had way too much fun playing it 😂

Grok Build with Grok 4.7 made this I basically asked it to build an interactive 3D journey from a human cell all the way down to DNA It handled the 3D scenes, animations, transitions and the entire experience The crazy part is I had to go out while it was still working, so I told it to keep going, record the finished result and send me the video on Telegram Then I just left it running... When it finished, Grok literally sent me the video on Telegram

Grok 4.7 Fast is now live in Grok Build and Cursor It’s the same Grok 4.7 model running on faster infrastructure, with roughly 2× the output speed • Available now in only Grok Build and Cursor • Costs 2× the standard Grok 4.7 token rates • Not included in Grok Build’s free tier • Not available on the public xAI API Same intelligence, much faster inference......at 2× the cost

Grok 4.7 isn't looking good it’s been 2 hours since model release and here’s what everyone is talking about: > better than 4.6 but only a small jump > uses way more tokens > frontend and ui are just not there > looks benchmark maxxed > after month of wait, this model isn’t looking good > same price as 4.6 and faster than sol and opus > good value for cursor and grok build users only from my testing, it’s just not there and still far behind frontier models. really disappointed tbh how’s your experience with 4.7 so far?

Grok Bot Summary: Grok 4.7 Model Card What it is - SpaceXAI / xAI’s latest model - Stronger at long, hard, agent-style work: coding, engineering, and office / knowledge tasks - Gets more done with fewer steps and fewer output tokens than many peer frontier models - Text in/out, plus image input - Extra training on anonymized Cursor workflow data for coding agents - Knowledge cutoff: June 2026 (some supplemental data through August 2026) Where you can use it now - SpaceXAI API - Grok Build (default model in their terminal coding agent) - Cursor (every plan) - Office add-ins for Word, PowerPoint, Excel - Model gateways (OpenRouter, Vercel, Cloudflare, Snowflake, Databricks, and others) - Consumer apps / Grok-in-X: coming later How it was trained - Pretrain on public + licensed + internal data - Longer supplemental training than 4.6 - SFT + RL on reasoning, agent harnesses, STEM, software, knowledge work, kernel work, web, and CAD Coding (big jump vs 4.6) - CursorBench 4.0: 46.3% (xhigh) / 43.9% (high) — long real Cursor sessions - DeepSWE v1.1: 71.0% (high) — near GPT-5.6 Sol’s 72.7% - Terminal-Bench 4.0: 38.0% (xhigh) vs 4.6’s 20.3% - FrontierSWE V2: 29.0% (xhigh) vs 4.6’s 25.3% - SWE-Marathon v1.1: 46.0% (high) vs 4.6’s 31.9% Knowledge work - Legal Agent Benchmark: 19.6% (xhigh) — leads the listed peers (4.6 was 15.8%) Engineering / physical world - EEBench (circuits / chip design): 66.0% (xhigh) vs 4.6’s 60.0% - CADGenBench: 44.4% (high) vs 4.6’s 40.9% — tops the listed peers Medical / biology (helpful, not autonomous doctor) - HealthBench Professional: 56.7% (xhigh) vs 4.6’s 48.5% - LatchBio Capabilities: 44.5% (xhigh) — real bio data analysis, not just quiz recall Cyber (small gains; safer in production) - Unrestricted capability probes are slightly up vs 4.6 - Best use case they emphasize: finding and fixing vulnerabilities, not end-to-end attacks - Production safeguards measured separately; harmful/dual-use compliance is low (good) Bio / chem safety - Below their FAIF dual-use safety thresholds - No dual-use bio capability increase vs 4.6; often lower on risky probes - Dual-use refusal on severity-5 BioUseBench: 91.4% - Bio refusal recall: 100%; chem: 99.9% Jailbreaks, safety, behavior - Standard jailbreak compliance: 0.01% (lower is better) - Child-safety compliance: 0.0% (no failures in that eval) - Dishonesty under pressure (MASK-Rectified): 0.00% - Sycophancy: 0.03% - Self-harm: refuses help for harm, still points people toward support Important limits - Not for autonomous high-stakes decisions in medicine, law, finance, or safety-critical systems without human experts - Bound by SpaceXAI Acceptable Use Policy and terms - They say they never silently downgrade the model or swap in a weaker one One-line takeaway Grok 4.7 is SpaceXAI’s strongest agent model yet — especially for long coding, terminal work, legal agent tasks, EE/CAD, and clinical communication — with tighter safety on dual-use bio and jailbreaks, available now in API, Cursor, Grok Build, and office add-ins. Grok 4.7 Model Card: