Apresentamos o Grok 4.6
Desenvolvido para agentes de longa duração e trabalhos interativos e visuais mais ambiciosos.
Announcements
Model Card do Grok 4.6
O model card oficial do Grok 4.6. Ele documenta os benchmarks de capacidade e as avaliações de salvaguardas do modelo.
Cursor firma parceria com a SpaceX para treinamento de modelos
A Cursor firmou parceria com a SpaceX para acelerar nossos esforços de treinamento de modelos.
Built for coding and knowledge work alike
Long-horizon problemsLong-horizon
problems
Grok 4.6 stays with complex tasks across many steps. It uses tools, checks its work, adjusts its approach, and keeps moving toward a finished result.
Ambitious engineeringAmbitious
engineering
Grok 4.6 works across large codebases and extended engineering projects. It researches unfamiliar systems, edits across files, runs tests, and verifies the result.
In-depth knowledge workIn-depth
knowledge work
Beyond code, Grok 4.6 works across documents, spreadsheets, PDFs, and other professional artifacts. Give it a multi-step project that requires research, analysis, and synthesis.
Interactive and visual workInteractive and
visual work
Grok 4.6 turns broad product ideas into working first versions. It can establish an application's structure and visual language, build the core interactions, and refine the result through feedback.
Built for more than software engineering
Grok 4.6 builds on Grok 4.5 with a focus on long-running agents and ambitious interactive and visual work. It handles projects that span research, analysis, implementation, and several rounds of refinement.
It matches GPT-5.6 Sol on the Artificial Analysis Intelligence Index, a composite of nine benchmarks. Cursor subscription plans for individuals and teams include significant usage of the model.
A strong foundation
We trained Grok 4.6 jointly with SpaceXAI. A longer supplemental training run used curated model-generated data for reasoning and advanced technical concepts, high-quality engineering data, and an improved optimizer and training recipe.
We then used Grok 4.5 to regenerate supervised fine-tuning trajectories across reasoning efforts, agent harnesses, STEM, software engineering, and knowledge work. Reinforcement learning covered general coding, knowledge work, kernel optimization, web development, computer-aided design, and other agentic environments.
Benchmarks
| 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 v1.1Extended | 61.3% | 56.6% | 60.6% | 64.9% |
| 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 LABVals | 15.8% | 12.9% | 2.5% | 11.3% |
Third-party model scores are the best self-reported or publicly available results.
(Above) Grok 4.6 results across agentic coding and knowledge work benchmarks.
Available everywhere you work
Grok 4.6 is available today in Cursor across desktop, web, iOS, CLI, and our SDK.
Interactive demo with multiple windows showing Cursor's AI-powered features. The interface is displayed over a subtle, solid brand background.
Set up Datadog APM for all API routes
Explored 12 files, 3 searches
I'll add the Datadog SDK, instrument all API routes with tracing, and set up error tracking.
Worked for 7m 42s
Processed screen recording
Done — here's the Datadog dashboard showing all instrumented routes.
Acme Labs
Summary
Integrated Datadog APM across all 24 API routes with error tracking and custom metrics.
Add a follow up...
This element contains an interactive demo for sighted users. It's a demonstration of Cursor integrated within Slack, showing AI-powered assistance inside team communication. The interface is displayed over a subtle, solid brand background.
FAQ
Grok 4.6 is a frontier model from Cursor and SpaceXAI for complex coding and knowledge work. It builds on Grok 4.5 with improved instruction following, long-horizon agentic work, and stronger first passes on interactive and visual projects.
Grok 4.6 is available today in Cursor across desktop, web, iOS, CLI, and our SDK. You can also build your own agents on top of Grok 4.6 with our SDK docs.
Grok 4.6 is part of the Cursor Models pool on individual and team plans, alongside Grok 4.5 and Composer 2.5. Standard on-demand usage is $2/M input, $0.50/M cached input, and $6/M output tokens. The Fast variant is $4/M input, $1/M cached input, and $12/M output tokens. We're offering 2x included usage for the first week. See the model docs for full details.