Llama 4 Maverick
by MetaRank #165Score 70.7
Llama 4 Maverick 17B Instruct (128E) is a high-capacity multimodal language model from Meta, built on a mixture-of-experts (MoE) architecture with 128 experts and 17 billion active parameters per forward...
性能概览
评分
70.7
排名
#165
24小时变化
+1
7天变化
-12
状态
stable置信度
high信号评分
| 信号 | 标准化 | 权重 | 贡献 | 新鲜度 |
|---|---|---|---|---|
Capabilities capability | 66.7 | 20% | 13.3 | 2026-09-22T15:12:02.311Z |
Benchmarks benchmark | 75.5 | 30% | 22.6 | 2026-09-22T15:12:02.311Z |
Pricing pricing_tier | 99.3 | 15% | 14.9 | 2026-09-22T15:12:02.311Z |
Context Window context_window | 95.5 | 10% | 9.6 | 2026-09-22T15:12:02.311Z |
Recency recency | 35.5 | 15% | 5.3 | 2026-09-22T15:12:02.311Z |
Output Capacity output_capacity | 67.4 | 10% | 6.7 | 2026-09-22T15:12:02.311Z |
主要驱动因素
positive
Benchmarks
positive
Pricing
$0.65/M output tokens
$0.65
positive
Capabilities
Supports vision, tools, JSON mode, streaming
4/7
positive
Context Window
1.0M token context window
1.0M
功能
| 功能 | 支持 |
|---|---|
| 视觉 | 是 |
| 推理 | 否 |
| JSON模式 | 是 |
| 流式输出 | 是 |
| 函数调用 | 是 |
| 网页搜索 | 否 |
价格
输入/百万令牌
$0.19
输出/百万令牌
$0.65
上下文窗口
1049K
最大输出
16K