Grok 4.7 Day-One Hands-On: Strong Word-of-Mouth for Autonomous Coding

On September 22, xAI released Grok 4.7, drawing heavy day-one hands-on sharing from developers with largely positive reception—most saw it as a major step up from Grok 4.6. Initially it was only available in Grok CLI, with the web version and Grok bot not yet live (vista8).

Confirmed

• Developer ScriptedAlchemy reported Grok 4.7 working autonomously on a single goal for over 70 hours, with much better attention to detail and effective use of the 500K-token context (m3)
• Developer Matt Shumer called 4.7 a very capable daily-driver model and a huge improvement over 4.6 (m2)
• Developer tetsuoai reported Grok 4.7 building a custom game engine and playable FPS in pure C (SDL2 and OpenGL) in about 7 hours, calling it his favorite coding model and recommending Grok Heavy mode with maxed tokens (m4, m5)
• One developer noted markedly better system-prompt adherence, pushing back on negative coverage citing poor public benchmarks, arguing benchmarks and 3D game demos don't reflect real work capability (m1)
• Developer DanielFarinax tested procedural watch exploded-view and HTML water-glass generation with solid results; he is also rebuilding his Grok-driven browser Xplor with 4.7, achieving things 4.6 couldn't, though he felt 3D tasks were weaker and still needed extreme prompting (m8, m14)
• Developer Baconbrix asked for a procedural game map based on Boca Chica, Texas; the model ran overnight and auto-added biomes, day-night cycles, fireflies and swamp-bubble particles, impressing him (m15)
• vista8 generated a playable Plants-vs-Zombies-style web game in one prompt, with zombies and items drawn in SVG in an "ugly-cute" style (m11, m12, m17)
• Users compiled 10 day-one extreme use cases, including a Boost Grok Bot productivity workflow, catching a $146,000 error for a business, Blender scenes, cloth simulation and robotic-arm animation, with commenters asking how to access the 500K-token context; CodeByPoonam also rounded up 7 community projects within 24 hours (m7, m10, m13, m16)

Unconfirmed

• A user posted a comparison video labeled Grok 4.7 vs 4.6 building open-world city games, showing 4.7 clearly stronger in complex 3D modeling, collision physics, kinematics and lighting—dubbed the "most fun benchmark"—but the version label is unverified (m6, m9), and it sits in tension with DanielFarinax's weaker 3D impressions
• DanielFarinax's account of rebuilding Xplor with 4.7 is self-reported and unverified (m14)

Why it matters

Day-one testing focused on real engineering scenarios—long autonomous work, ultra-long context use, and low-level systems programming—rather than benchmark scores, offering third-party evidence for evaluating frontier-model productivity. If 500K-token context effectiveness and 70-hour autonomous runs are further validated, they would be an important signal about the capability boundaries of AI coding agents.

2026-09-22 ~ 2026-09-23 · 17 related posts

Full story(9 episodes)→

• Episode 1: Grok 4.7 Briefly Surfaces in OpenCode Zen Directory, Fueling Launch Rumors(2026-09-21, 5 posts)
• Episode 2: xAI launches Grok 4.7 with major benchmark gains at unchanged price and speed(2026-09-21, 59 posts)
• Episode 3: Grok 4.7 Day-One Hands-On: Strong Word-of-Mouth for Autonomous Coding(2026-09-22, 17 posts)
• Episode 4: Grok 4.7 Launches: Big Benchmark Gains but Doubled Token Use Sparks Cost Debate(2026-09-22, 25 posts)
• Episode 5: Grok 4.7 Reportedly Closes Gap With Top Frontier Models at Lower Cost(2026-09-22, 2 posts)
• Episode 6: Open-Source Grok 4.7 Editor Extension Passes 140K Installs(2026-09-22, 3 posts)
• Episode 7: Bug Hunt Bench: GPT-6 Astra Tops 105-Bug Fixing Test, MiMo Stuns on Value(2026-09-22, 18 posts)
• Episode 8: Grok 4.7 Meets Xiaomi MiMo-V2.6 in a Same-Night Price War(2026-09-22, 5 posts)
• Episode 9: xAI Ships Three Grok Frontier Models in Nine Weeks(2026-09-22, 2 posts)

• [source] Grok 4.7 goes live in Grok CLI, with web and bot access still missing — vista8 · 2026-09-22
• Grok 4.7 First Playtest: SVG-Drawn Zombies in a Vibe-Coded PvZ Clone — vista8 · 2026-09-22
• Grok 4.7 comparison clip shows major gains in 3D modeling and game physics over 4.6 — belce_dogru · 2026-09-22
• Grok 4.7 generates a playable Plants vs Zombies from a single prompt — vista8 · 2026-09-22
• [source] Matt Shumer on Grok 4.7: a great daily driver and huge step up from 4.6 — elonmusk · 2026-09-22
• Grok 4.7 vs 4.6 Compared by Building an Open World City Game — "Best Benchmark Ever" — NicoVerderosa · 2026-09-22
• Grok 4.7 tested: HTML water glass and procedural watch exploded views impress — Daniel_Farinax · 2026-09-22
• Dev is building a game engine in C with Grok 4.7: "really good" — elonmusk · 2026-09-22
• [source] Dev review: Grok 4.7 ran autonomously for 70+ hours, 500k context is a game changer — elonmusk · 2026-09-22
• Grok 4.7 launch day: 10 wild demos from Blender scenes to $146K error catches — minchoi · 2026-09-22
• Dev's early take: Grok 4.7 excels at complex coding but falters on 3D — Daniel_Farinax · 2026-09-22
• Day 1 with Grok 4.7: strict system-prompt adherence and visible gains over 4.5 in real coding work — elonmusk · 2026-09-22
• Grok 4.7 day one: users find $146K bug, build Blender scenes in 10 wild demos — FinanceYF5 · 2026-09-22
• Dev impressed by Grok 4.7: it ran all night to build a procedural game map of Boca Chica — Baconbrix · 2026-09-22
• Grok 4.7 builds a custom game engine and playable FPS in C in about 7 hours — tetsuoai · 2026-09-22
• Grok 4.7 drops, users already catch $146K in errors and build Blender scenes — socialwithaayan · 2026-09-22
• 7 wild builds from Grok 4.7's first 24 hours — CodeByPoonam · 2026-09-23