SpaceXAI has launched Grok 4.6, matching ChatGPT 5.6 Sol in several AI benchmarks
SpaceXAI has launched Grok 4.6, the newest version of its widely used artificial intelligence model. This update is now accessible across platforms such as Cursor, Grok Build, OpenRouter, Vercel, Cloudflare, and through the API.
Focusing on extended agent capabilities, Grok 4.6 is tailored for complex, multistep tasks. Users can maintain progress whether conducting in-depth research, analyzing data, navigating source code, or transforming ideas into complete digital outputs. The model stands out for its improved support in turning broad product ideas into functional first versions. It handles research in new subject areas, organizes applications, manages core implementation, and iteratively refines results using user feedback.
Following functional upgrades, Grok 4.6 also boosts productivity in interactive and visual project development. Compared to Grok 4.5, it yields better first drafts by developing the application’s structure and visual language in a single step.
These advances are supported by Grok 4.6 achieving parity with GPT-5.6 Sol on the Artificial Analysis Intelligence Index, comprised of nine coding and knowledge-based benchmarks. To match its expanded scope, Grok 4.6 integrates upgraded and recalibrated safeguards, aligning safety measures with its enhanced technical abilities.
- AI Chatbot
- Freemium
- Proprietary
Related news
External links
- Introducing Grok 4.6SpaceXAI • Official source
- SpaceXAI releases Grok 4.6, claiming GPT-5.6 Sol and Claude Fable 5-level intelligence9to5Mac
- SpaceXAI's Grok 4.6 model beats OpenAI's GPT-5.6 Sol in several AI benchmarksNeowin
- SpaceXAI debuts Grok 4.6, overtaking Kimi K3's performance and matching GPT-5.6 Sol for world's third best on Artificial AnalysisVentureBeat
- SpaceXAI releases flagship Grok 4.6 model with advanced reasoning capabilitiesSiliconANGLE
- SpaceXAI's Grok 4.6 matches OpenAI's best model and undercuts it on priceThe Decoder
Comments
Elon is hyping Grok 4.7 so hard, can't wait for it.
The Intelligence Index is quite outdated, and outside it, on coding and agentic benchmarks, Grok is doing as good as Kimi (the latter being 30% cheaper). As for visual and design tasks, Grok fells far beyond Kimi, and even beyond Meta's no-one-care Muse Spark model.
OpenAI has launched 5.6 this summer (because of much lower use of AI during summer holidays) as the last model built on its legacy architecture, and is still hyping around "Astra" (its 10T parameters model with new architecture) from time to time, which could be announced as ChatGPT 5.7 in a month, that may not be a game changer but would be better than 5.6 on every level.
And then ChatGPT 6 is planned somewhere in 2027, even if it's hard nowadays to separate real breakthroughs from Altman's hallucinations about self-improving models, AGI and billion dollar IPO.