xAI/Grok 4.6
xAI’s Most Intelligent Model
A 500K-token context, stronger long-running agents, and top-tier coding — built for ambitious interactive and visual builds. Put it to work in your browser, no setup required.
Model preset to Grok 4.6 · Sign up to unlock more models
Trending Grok 4.6 use cases
The hottest practices from the X community — what people are actually building with Grok 4.6: games, 3D worlds, simulations, renders and head-to-head model comparisons.
Grok 4.6 vs Kimi K3 on the same prompt at highest reasoning — Grok finished all the tests in about 20 minutes for $15, while Kimi took 40 minutes for $12.
Build with Grok 4.6Grok 4.6 xhigh vs GPT-5.6 Sol Max on one prompt — Grok finished in about 23 minutes ($21.18) against Sol’s 33 minutes ($22.53).
Build with Grok 4.6One prompt, ~22 minutes of autonomous work: Grok 4.6 produced a rich scene with custom shaders, a minimap, and live time-of-day changes.
Build with Grok 4.6Pitted Grok 4.6 (Grok CLI) against Kimi K3 and DeepSeek V4 Pro on the same Three.js 3D-modeling and spatial-reasoning prompt to compare their output.
Build with Grok 4.6Built a retro Mario-style platformer with Grok 4.6 (Grok Code) vs Kimi K3 (Kimi CLI) — Grok was faster at ~11 minutes, though Kimi’s output edged ahead.
Build with Grok 4.6Ran FlappyBench across three frontier models — Grok 4.6 topped the group at 9.5/10 for just $0.095, ahead of GPT-5.6 Sol and Opus 5.
Build with Grok 4.6Ran a recurring NYC city-render prompt on Grok 4.6 — a huge jump over 4.5, with real density, lit windows, and an actual skyline.
Build with Grok 4.6A one-shot Three.js render of the opening paragraph of The Lord of the Rings — Grok 4.6 High even pulled in Kokoro-82M for audio.
Build with Grok 4.6One-shotted a real-time Gargantua black-hole simulation with Grok 4.6 xhigh — a browser Schwarzschild raytracer with lensing, a Doppler disk, and a photon ring.
Build with Grok 4.6Created an atmospheric “Moonlit Balcony” interactive scene with Grok 4.6.
Build with Grok 4.6Created a “Rainy Café Window” interactive scene with Grok 4.6.
Build with Grok 4.6Recreated a playable Super Mario from a single Grok 4.6 prompt.
Build with Grok 4.6Grok 4.6 wrote its own prompt and generated a video from it.
Build with Grok 4.6What is Grok 4.6?
Grok 4.6 is xAI’s most intelligent large language model, released on August 12, 2026 as the successor to Grok 4.5. It ties GPT-5.6 Sol at 61 on the Artificial Analysis Intelligence Index — one point behind Fable 5 at 62 — and is tuned for long-running agents and more ambitious interactive and visual work — holding focus across complex, multi-step research, analysis, and codebase tasks.
Grok 4.6 ships with a 500,000-token context window, accepts text and image input, and has a February 2026 knowledge cutoff. On Happycapy it isn’t just a chat box: it powers an agent inside a secure cloud sandbox that can write code, run it, render, drive tools, and deliver finished work — all from a single prompt, with no local setup.
How does Grok 4.6 perform on benchmarks?
Grok 4.6 compared with Fable 5, GPT-5.6 Sol, and Grok 4.5 across the Artificial Analysis Intelligence Index, GDPVal-AA, DeepSWE 1.1, CursorBench 3.2, and FrontierCode 1.1. Switch tabs to see each eval.
Competitor figures are drawn from the respective developers’ published system cards or benchmark leaderboards.
Source: xAI — Grok 4.6
Evals
| Grok 4.6 High | Grok 4.5 High | GPT-5.6 Sol Max | Fable 5 Max | |
|---|---|---|---|---|
| AA Intelligence Index | 61 | 56 | 61 | 62 |
| GDPVal-AA v2 | 1753 | 1526 | 1728 | 1741 |
| CursorBench v3.2 | 69.9% | 66.7% | 67.2% | 70.5% |
| DeepSWE v1.1 | 65.9% | 54% | 73% | 70% |
| FrontierCode v1.1 (Extended) | 61.3% | 56.6% | 60.6% | 63.6% |
| APEX-Agents | 57.5% | 47.1% | 56.7% | 59.2% |
| Terminal-Bench v3.0 | 26% | 15.7% | 34.6% | 34.1% |
| APEX-SWE | 56.4% | 53.6% | — | 58.8% |
| AA-Briefcase | 1577 | 1313 | 1502 | 1574 |
| Harvey LAB (Vals) | 15.8% | 12.9% | 2.5% | 11.3% |
Best score per evaluation in bold. Third-party model scores are the best of self-reported or publicly available results.
What Grok 4.6 is best at
Long-running agents
Extended, multi-step tasks — research, analysis, and codebase work — where it self-tests and verifies its own progress instead of losing the thread.
Ambitious interactive & visual work
Produces stronger first passes on visual and interactive projects, and is better at establishing an application’s structure in a single iteration.
Agentic coding
Top-tier coding scores (CursorBench 69.9%, DeepSWE 65.9%, FrontierCode 61.3%) for autonomous, multi-file build loops.
Frontier reasoning at speed
Ties GPT-5.6 Sol at 61 on the Artificial Analysis Intelligence Index — just behind Fable 5 at 62 — while staying fast and token-efficient, with an even faster premium variant available.
Who is Grok 4.6 for?
Developers & agent builders
Anyone running autonomous, multi-file coding loops who wants frontier-class output that holds up over long tasks.
Product & app builders
Makers turning a single prompt into an interactive app or tool, leaning on strong first-pass structure.
Startups & solo builders
Shipping MVPs and demos where intelligence, speed, and token cost all matter at once.
Creative technologists
People bridging code and visuals — generative graphics, interactive scenes, and data-heavy interfaces.
Grok 4.6 FAQ
What’s new in Grok 4.6 compared to Grok 4.5?+
Grok 4.6 is xAI’s newer, more intelligent model, released August 12, 2026. It went through extended supplemental training on curated model-generated data for reasoning and technical concepts, with SFT trajectories regenerated using Grok 4.5 and improved filtering. In practice it shows stronger self-testing and verification on extended tasks, produces better first passes on visual and interactive projects, and is better at establishing application structure in a single iteration.
How intelligent is Grok 4.6?+
On the Artificial Analysis Intelligence Index, Grok 4.6 scores 61 — tying GPT-5.6 Sol and sitting just one point behind Fable 5 at 62. It also posts strong coding and agentic results: CursorBench v3.2 69.9%, DeepSWE v1.1 65.9%, FrontierCode v1.1 61.3%, and GDPVal-AA v2 1753.
How large is Grok 4.6’s context window?+
Grok 4.6 has a 500,000-token context window and accepts both text and image input, with text-only output. Its knowledge cutoff is February 2026.
How much does Grok 4.6 cost?+
xAI lists Grok 4.6 at $2 per million input tokens and $6 per million output tokens (cached input is $0.50 per million); prompts of 200K+ tokens are billed at the higher $4 / $12 tier, and a faster premium variant costs double. On Happycapy, Grok 4.6 is a premium model on paid plans — you can create a free account to explore the platform first, then upgrade to run it.
Can Grok 4.6 build a complete app on its own?+
Yes — that is a core use case. It is tuned for long-running agents that plan, write code, run it, and iterate autonomously, and it is notably better at establishing an app’s structure on the first pass. On Happycapy it runs inside a secure cloud sandbox that turns a single prompt into finished, runnable work.
Where can I use Grok 4.6?+
At launch xAI made Grok 4.6 available across Cursor, Grok, the API console, OpenRouter, Vercel, and Cloudflare. On Happycapy it runs as the engine behind an in-browser agent, so you get an agentic build workflow with no API keys or local setup.
How do I run Grok 4.6 on Happycapy?+
Type your task into the prompt box on this page — the model is already set to Grok 4.6 — and hit send. Happycapy spins up a secure cloud sandbox and lets Grok 4.6 build, code, or render the task end-to-end, then delivers the result. No installs or API keys required.
Build with Grok 4.6 now
Create a free account to explore Happycapy, then upgrade to put xAI’s most intelligent model to work on real tasks — right in your browser, no setup needed.
Get started for free