ModelsSelected73°9月7日 12:15

iFlytek Spark-X2.5 model released

科大讯飞星火 X2.5 模型正式发布：293B-A30B MoE，全国产平台训练

IT之家

Blog/Media

Why we picked it

Priced competitively against alternatives, Spark-X2.5 offers strong multilingual support and is accessible through iFlytek's MaaS platform.

iFlytek launched Spark-X2.5 on 7 Sep 2024.

Body

The model uses MoE architecture with 293B-A30 parameters and supports 256K context.It improves code and agent capabilities while covering 200+ languages.

The model is available on iFlytek's MaaS platform with pricing at RMB 1.6/million tokens for input and RMB 6/million tokens for output.

Original · IT之家

科大讯飞星火 X2.5 模型正式发布：293B-A30B MoE，全国产平台训练

IT之家 9 月 7 日消息，科大讯飞 (iFlytek) 今日正式发布星火 Spark-X2.5 多语言文本生成模型。该模型基于全国产平台训练，采用 MoE 框架，模型参数 293B-A30，支持 256K 上下文。星火 X2.5 全面提升代码和智能体能力，同时覆盖 200 余种语言，在语言理解、答题、推理等通用任务上保持优异的效果。这款模型目前已在讯飞星辰 MaaS 平台上线，定价如下：输入：1.6 元人民币 / 百万词元 (Token)；输入缓存命中：0.24 元人民币 / 百万词元；输出：6 元人民币 / 百万词元。

Industry Coordinates & Source Verification

This report is categorized under AITOP's "Models" desk, focusing on China's AI industry and domestic foundation model ecosystem. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows and US-China foundation model benchmarks and commercialization comparison. Key entities and technical topics under continuous tracking include iFlytek, Spark-X2.5, MoE, multimodal.

Source attribution: curated from IT之家.

#iFlytek #Spark-X2.5 #MoE #multimodal #open-source models

Keep reading

Recent stories with the same tag or section

Models Builders13:56

Chinese content科大讯飞发布Spark X2.5 AI模型

讯飞新出的Spark X2.5专攻编程和智能体，还开源了小模型支持长上下文。

#Spark X2.5 #iFlytek #编程助手 #开源模型

Event

Tech in Asia@Aiko Gao IshidaSource

Models US vs ChinaOfficial18:54

Chinese contentGoogle 发布 Gemini 4 Hydrogen：速度较 Gemini 3.8 Flash 提升 2.5 倍

Google 新出的 Gemini 4 Hydrogen 主打快，比 Gemini 3.8 Flash 快 2.5 倍，还带 1M 上下文和 4 档思考模式。跑 agent 或批量任务的开发者可以关注下这个价格和 Terminal-Bench 成绩，算算换过来省不省。

#Gemini 4 Hydrogen #Google #推理模型 #智能体

Event

Primary sourcemarktechpost@Michal SutterSource

Models US vs ChinaSelected17:33

Tencent Hunyuan Hy4 Ranks #3 on Artificial Analysis Cyber Index

Cost efficiency rather than raw leaderboard position is the real story: matching near-frontier security performance at roughly 1/11th the cost of Grok 4.7 (xhigh) changes the calculus for teams running vulnerability-scanning agents at volume. Security engineering teams and platform builders evaluating agentic models for code audit pipelines can use the per-task cost and token-consumption figures directly in procurement comparisons. Against DeepSeek V4.1 Flash 1017 (max), the tradeoff is higher token usage for a slight accuracy edge.

#Tencent Hunyuan #Hy4 #DeepSeek #Kimi K3.1 Pro