xAIAPI
Grok 4.6 API
8/10Efficiency Score
$2/1M tokens
Pricing Details
Input Tokens$2/1M
Output Tokens$6/1M
Context Window500K
Rate LimitAPI 默认
Features
Pay per use
500K context window
Knowledge cutoff February 1, 2026
Configurable reasoning (effort levels)
Native tool calling and agent support
AA Intelligence Index 61 (matches GPT-5.6 Sol)
DeepSWE 1.1 65.9% (vs Grok 4.5 54%)
CursorBench 3.2 69.9% (vs Grok 4.5 66.7%)
FrontierCode 1.1 61.3% (vs Grok 4.5 56.6%)
Terminal-Bench 3.0 26% (vs Grok 4.5 15.7%)
Harvey LAB (Vals) 15.8% (vs Grok 4.5 12.9%)
🔥 8/12 xAI 发布 Grok 4.6（500K 上下文，输入 $2/输出 $6 per 1M tokens，Fast variant 2x = $4/$12；AA Intelligence Index 61 = GPT-5.6 Sol 持平，落后 Opus 5/Fable 5；已集成 Cursor/Grok Build/OpenRouter/Vercel/Cloudflare，首周 Cursor 享 2 倍用量）
🔥 Grok 4.6 强项：长任务 Agent 持久工作（research / codebase / polished app）；视觉与交互项目一次成型；自测试和验证能力更强；'first pass' 视觉质量显著优于 4.5
Best For
developersagentslong_running_tasks