xAI Launches Flagship Grok 4.6, Rivaling GPT-5.6 and Claude 5 at Lower Cost
智东西 · wechat · 2026-08-13
Elon Musk announced the launch of Grok 4.6, a new flagship model emphasizing intelligence, speed, and cost-effectiveness. It offers significantly optimized performance at the same price point, with overall capabilities rivaling GPT-5.6 and Claude Fable 5 Max.
Grok 4.6 is now available on Cursor, Grok Build, and other platforms like OpenRouter, featuring a 2x usage bonus in the first week. Pricing is set at $2 per million input tokens and $6 per million output tokens. It scored 1618 on the CodeArena WebDev benchmark.
The model underwent extended training using filtered synthetic data and regenerated SFT trajectories from Grok 4.5. It excels at multi-step complex tasks, effectively turning broad product concepts into runnable applications, and demonstrates emerging self-testing and verification behaviors.
Related event: xAI Launches Grok 4.6: Top Benchmark Performance at Low Cost(27 posts)→
More from Models
- Qwen 1-bit Quantization Shrinks Model to 397GB, a 91% Reduction — danielhanchen · 2026-08-13
- Users Complain About Claude's Unstoppable Chain-of-Thought Output — GuyHachmon · 2026-08-13
- OpenAI's gpt-live-1 Achieves Near-Perfect Conversational Turn Detection — pbbakkum · 2026-08-13
- xAI Launches Grok 4.6 Across Cursor, API with 2x Token Promo — aman_madaan · 2026-08-13
- xAI Exec Hints at Grok 4.5: Focused on Coding Agents — aman_madaan · 2026-08-13
- Reddit Speculates: Is Grok 4.6 a Fine-tune of Kimi K3? — robertpro01 · 2026-08-13