Meta's latest class of model (Llama 3) launched with a variety of sizes & flavors. This 70B instruct-tuned version was optimized for high quality dialogue usecases. It has demonstrated strong...
| 信号 | 强度 | 权重 | 影响 |
|---|---|---|---|
| Benchmarksjust now | 60 | 30% | +18.0 |
| Pricingjust now | 99 | 15% | +14.9 |
| Output Capacityjust now | 65 | 10% | +6.5 |
| Context Windowjust now | 62 | 10% | +6.2 |
| Capabilitiesjust now | 17 | 20% | +3.3 |
| Recencyjust now | 0 | 15% | +0.0 |
LMSYS Arena Elo
1275
Percentile
62.5
Weight
30%
把当前模型放回同一服务商最近的发布节奏中查看。
Llama Guard 4 12B
coding
Llama 4 Maverick
coding
Llama 4 Scout
coding
Llama Guard 3 8B
coding
Llama 3.3 70B Instruct (free)
coding
Llama 3.3 70B Instruct
coding
Llama 3.2 3B Instruct (free)
coding
Llama 3.2 3B Instruct
coding
社区和从业者反馈在基准测试和价格之上增加了真实世界的信号。
Share your experience with Llama 3 70B Instruct and help the community make better decisions.
LMSYS Arena Elo
1275
Percentile
62.5
Weight
30%
成本估算器
每月比类别平均节省$39.85