Open source Kev model released by Cognition team
For developers, Kev offers a low-cost, open-source alternative to proprietary decision models like Jev, with a small footprint and fast inference times.
Latest
Every published item, in order · 341 items
Sep 21
Sep 20
Alibaba releases Qwen3.8-LiveTranslate model for real-time speech translation
For users needing real-time translation, this model reduces latency by 0.5 seconds compared to the previous version, improving the user experience.
Qwen3.8-LiveTranslate updates its architecture
This update improves the efficiency and speed of the Qwen3.8-LiveTranslate model by optimizing its internal architecture.
Sep 19
Qwen3.8-LiveTranslate model released
For users needing real-time translation, this new model from Alibaba offers lower lag and better speaker separation compared to previous versions.
Bespoke Labs releases open-source Nimble model
This matters to developers because it provides a new, open-source approach to fine-tuning models for structured decision-making, offering a concrete alternative to the closed-source Jev.
Sep 18
Qwen3.8-Omni-Flash, Qwen's first omni-modal model with agentic capabilities
For users needing to process long-form audio-video content, Qwen3.8-Omni-Flash is more affordable than its predecessor, with a 89% reduction in video input costs.
Tencent WeChat visual team opens WeVisDoc document parser
For users needing to process documents in Chinese and English, WeVisDoc offers a free, open-source alternative to commercial models like HunyuanOCR-1.5, with demonstrated improvements in handling degraded or real-world images.
Sep 12
Qwen3.8-27B model runs on Cerebras for faster inference
The Qwen3.8-27B model is now available on Cerebras hardware, offering faster inference speeds compared to its previous deployment.
Sep 11
Sep 10
Chinese contentAgent 通过率提升 19%，模型参数未动
Adaption AI 证明 Agent 表现差往往不是模型问题，而是 Harness 设计问题。优化 Harness 能大幅提升性能，还能降低成本。
Sep 9
KVMem improves inference efficiency
KVMem enables local deployment of large models with extended context on consumer hardware, making long-running agents accessible without expensive cloud infrastructure.
Sep 5
Tencent environment evolution paper
The method improves RL performance significantly on Qwen models, relevant for developers working with reinforcement learning in ASEAN.
Sep 4
Chinese contentThomson Reuters 推出大模型 Thomson
Thomson Reuters 花费 4000 万美元训练自己的大模型，用 175 年专业数据，性能接近 Claude Opus 4.8。
Fable 5.1 (Max) tops Code Arena rankings
Fable 5.1 (Max) offers top-tier performance at $40/MToken, making it a competitive option for developers in the region seeking advanced AI coding assistance.
Sep 3
Chinese contentPerplexity 开源 Mac 本地推理引擎 Lily
Perplexity 为 Mac 开发了专用推理引擎 Lily，比 MLX-LM 快 23%-35%，专为 Qwen3.6-35B-A3B 和 Apple Silicon 优化。
Chinese contentPerplexity开源Lily本地推理引擎
Perplexity开源了专为Apple Silicon优化的Lily推理引擎，让Qwen3.6-35B-A3B在本地运行更高效。
Chinese contentQwen3.6-35B-A3B 在 MacBook Pro 上性能测试
Qwen3.6-35B-A3B 在苹果笔记本上跑得比 MLX-LM 还快，吞吐量提升超 20%，质量不打折。
Sep 2
Chinese contentQwen3.8-Max-0902登顶CodeArena WebDev排行榜
阿里Qwen3.8-Max-0902以1691分登顶CodeArena WebDev排行榜，比前代模型高出22分，价格仅5美元/百万token。
Qwen3.8-Max-0902 tops Code Arena
Qwen3.8-Max-0902 offers top coding performance at competitive pricing compared to alternatives like Claude Opus 5 and Kimi K3.
Qwen3.8-Max-0902 tops Code Arena leaderboard
Qwen3.8-Max-0902 offers top coding performance at a competitive price point, making it a strong alternative to more expensive models like Claude Opus 5 and Kimi K3.
Chinese contentQwen3.8-Max-0902登顶Code Arena: WebDev
阿里Qwen3.8-Max-0902在代码基准测试中超越Claude和Kimi，价格实惠，多项类别领先。
Chinese contentShopify基于Qwen3.5-0.8B微调模型击败GPT 5.6-sol xhigh
Shopify用0.8B小模型微调后打败了GPT 5.6-sol xhigh，小模型专用训练有奇效。
Chinese contentQwen团队发布电商智能体基准测试
Qwen团队测试了18个模型在365天电商运营中的表现，GPT-5.6 Sol盈利最多但效率不高，Qwen3.8-Max-Preview在开源模型中领先。
Sep 1
Chinese contentLoopArena：模型作为管理者的可测量基准
LoopArena 基准首次单独测量模型作为'管理者'的能力，发现 GPT-5.5 在长程任务中表现最佳，但成功率仍不足 25%。
Chinese contentQwen3.8-Flash-Next在Agent Arena排名第7
阿里Qwen3.8-Flash-Next在Agent Arena排名第7，比前代模型提升2.4%，任务完成率高达12.3%。
Qwen3.8-Flash-Next model ranks 24th
Qwen3.8-Flash-Next ranks among top open-source models, offering strong performance for developers in the region.
Aug 31
Aug 30
Aug 28
OpenCode Go launches Qwen3.8-Flash
Qwen3.8-Flash provides developers in ASEAN with a powerful multimodal coding assistant at competitive pricing.
Agent skill library malware propagation
A counter-prompt reduces poisoning to 6.7% without affecting task completion, offering developers a practical defense against this vulnerability.
Aug 27
Qwen3.8-Flash on OpenRouter
Available through OpenRouter API, offering developers in ASEAN access to Alibaba's multimodal model for coding and visual tasks at competitive pricing.