Fireworks AI：路由让 113 个编码任务解题率达 97.6%
Fireworks AI 测了 18 个模型：把任务路由给最擅长的那个，比单用最强模型更准也更省。
Fireworks AI 在 DeepSWE 上用 113 个真实编码任务评测了 18 个模型。单一最强模型的解题率为 74.
1%，每任务成本 6.52 美元。
把每个任务路由给最擅长的模型后，整体解题率升至 97.6%，每任务成本降至 1.
88 美元。Fireworks AI 在博客中称，模型路由是下一个前沿方向。
We ran 18 models across 113 real coding tasks on DeepSWE, then went back and asked a simple question: what if every task had routed to the model that handled it best? Answer: 97.6% solve rate at $1.88 per task, versus the best model at 74.1% at $6.52. The next frontier is a router. Full analysis: fireworks.ai/blog/the-front…📊 1 ⚡
This report is categorized under AITOP's "Papers" desk, focusing on US frontier AI labs and Silicon Valley technology leaders. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows. Key entities and technical topics under continuous tracking include Fireworks AI, DeepSWE, 模型路由, 模型评测.
Source attribution: curated from Fireworks AI (@FireworksAI_HQ).
