リリース2026/05
生成速度0 トークン/秒
初動レイテンシ (TTFT)0.00s
入力料金$0/1M
出力料金$0/1M
ランキング
レビュー
JT-35B-Flash is a solid choice for heavy agentic and reasoning tasks, holding the #1 rank in AI Agent performance. However, it is not suitable for general chatbots or conversational applications. It is best suited for teams in China and Southeast Asia needing a lightweight, low-cost model with strong automation capabilities.
長所
- Exceptional AI agent performance (ranked #1): excels at workflow automation and executing complex agentic commands
- Strong reasoning (ranked #66): handles logic, mathematics, and structured analysis well
- Lightweight 35B model: offers flexible deployment and low inference costs
短所
- Weak general chat (ranked #81): struggles with open-ended conversation, long context, and creative content generation
- Unclear context length and response latency compared to peers in its class
- Chinese ecosystem reliance: limited accessibility and integration outside the region
ユースケース
Autonomous agents & workflow automation: orchestrating automated business tasksData analysis & reasoning: processing complex data and solving logical problemsBackend reasoning engine: integrating into systems that require heavy logic without conversational polish
ガイド・動画
JT-35B-Flash is a language model developed by China Mobile, accessible via the MoMA platform. It supports text processing and is well-suited for coding, mathematics, translation, and data processing tasks. You can access it through MoMA's API gateway or via third-party API providers integrated with China Mobile.