Llama 4 Maverick is a natively multimodal model capable of processing both text and images. It features a 17 billion active parameter mixture-of-experts (MoE) architecture with 128 experts, supporting a wide range of multimodal tasks such as conversational interaction, image analysis, and code generation. The model includes a 1 million token context window.
发布日期2025年4月5日
参数规模400B
上下文长度1.0M
许可证Llama 4 Community License Agreement
知识截止—
Benchmarks
评测成绩
| 评测基准 | 类别 | 分数 | 来源 |
|---|---|---|---|
| DocVQA | vision multimodal | 94.4 | 来源 |
| MGSM | math reasoning | 92.3 | 来源 |
| ChartQA | reasoning vision multimodal | 90.0 | 来源 |
| MMLU | general reasoning language math | 85.5 | 来源 |
| MMLU-Pro | language reasoning math general | 80.5 | 来源 |
| MBPP | reasoning general | 77.6 | 来源 |
| MathVista | math vision multimodal | 73.7 | 来源 |
| MMMU | multimodal reasoning general | 73.4 | 来源 |
| GPQA | reasoning general | 69.8 | 来源 |
| MATH | math reasoning | 61.2 | 来源 |
| MMMU-Pro | vision multimodal reasoning general | 59.6 | 来源 |
| LiveCodeBench | reasoning general code | 43.4 | 来源 |
| TydiQA | language reasoning | 31.7 | 来源 |
Pricing
API 价格对比
| 服务商 | 输入价 | 输出价 | 上下文 | 吞吐(tok/s) | 延迟(s) | 函数调用 | 代码执行 | 联网搜索 |
|---|---|---|---|---|---|---|---|---|
| Cortecs | $0.12 | $0.60 | 1.0M | — | — | ✓ | ✗ | ✗ |
| Helicone | $0.15 | $0.60 | 131K | — | — | ✓ | ✗ | ✗ |
| DigitalOcean | $0.25 | $0.87 | 1.0M | — | — | ✓ | ✗ | ✗ |
| Neon | $0.50 | $1.50 | 1.0M | — | — | ✓ | ✗ | ✗ |
价格单位：美元/百万 token，数据来自社区整理，仅供参考。