Chinese contentDeepSeek 发布开源模型 DeepSeek-V4.1-Flash-1017
DeepSeek 新出的 V4.1-Flash-1017 是开源的，DeepSWE 1.1 跑分 77%，把 GPT-5.6 Sol 和 Opus 5 都比下去了，价格还便宜。
Latest
Every published item, in order · 503 items
10月17日
Chinese contentDeepSeek 发布 V4.1-Flash-1017 并下调 API 价格
DeepSeek 周末更新的 V4.1-Flash-1017，性能更强还降价，搜索直接做成 API，5 元一千次，Agent 开发者可以马上试试。
Chinese contentDeepSeek V4.1 Flash 1017 发布，强化从零构建完整项目能力
DeepSeek 周末悄悄更新了 V4.1 Flash 1017，主打一句描述就能生成完整游戏或 3D 项目，做独立开发 or 玩编程的可以试试。
Chinese contentV4.1 Flash 1017 通过自反馈 RL 强化从零搭建完整项目能力
V4.1 Flash 1017 更新了，用自反馈 RL 让模型能从零做出完整游戏和 3D 项目，搜索功能被网友点赞。
Chinese contentDeepSeek-V4.1-Flash 小幅更新：推理、编程与多模态基准提升
DeepSeek 给 V4.1-Flash 做了次小更新，推理、编程、看图都变强，跑任务还更快更省 token，开源党可以关注下。
Chinese contentDeepSeek V4.1 Flash 更新至 1017 版本，网页端与 API 同步替换
DeepSeek 的 V4.1 Flash 换了新版本号 1017，网页和 API 都自动切过去了，不用重新配置就能用上。
IT之家Source
Chinese content博主点赞 DeepSeek V4.1 的多视角智能体表现
有博主晒 DeepSeek V4.1 跑多视角智能体的效果，说分歧视角输出很有意思，想看 V4.1 实际表现可以瞄一眼。
Chinese contentDeepSeek V4.1F-1017 疑似曝光：5 分钟生成前端界面，对比力压 Opus
有人拿没官宣的 DeepSeek V4.1F-1017 做前端，5 分钟出活，还敢说比 Opus 强，看看截图自己判断。
Chinese contentDeepSeek-V4.1-Flash-1017 疑似就是 DeepSeek-V4.2-Flash
有人发现 DeepSeek 这两个版本号可能其实是同一个模型，用 DeepSeek 的朋友可以看看自己的 API 版本对应关系。
Chinese contentDeepSeek V4.1-Flash-1017 疑似上线：deepseek-flash 的 system_fingerprint 已变更
DeepSeek 的 flash 模型指纹变了，疑似悄悄上了 V4.1-Flash-1017，有人说变聪明也更简洁了，可以自己试试对比。
Chinese contentMeta Muse Spark Open 测试速报：复杂算法尚可，空间理解拉胯
有人实测了 Meta 新开源的 Muse Spark Open，结论是打不过 DeepSeek-v4.1-flash，还特能幻觉，想本地部署的先看看再决定。
Chinese contentMeta 发布开源权重模型 Muse Spark Open：816B 参数，登顶美国开源权重榜首
Meta 放了个 816B 的开源权重模型 Muse Spark Open，美国开源第一，1M 上下文还能塞进两台 Mac Studio 自托管，不过跑智能体任务不如 GLM-5.3。
10月16日
Chinese content微软完成 MAI-Base-2 预训练，2.8T 参数强于所有参测开源权重模型
微软自研基模 MAI-Base-2 训完了，2.8T 总参数，算力效率是 DeepSeek-V4-Pro-Base 的 4~12 倍，还放出了训练细节。
Chinese contentNVIDIA Vera Rubin NVL72 首批基准成绩公布
NVIDIA 新一代 Vera Rubin NVL72 的跑分出来了，跑 DeepSeek V4 Pro 每兆瓦吞吐量是上一代的 5.7 倍，还刷出 906 tok/s 的单用户最快速度。
Chinese contentMeta 开源 Muse Spark Open：816B 参数、1M 上下文、支持全模态输入
Meta 开源了 816B 的 Muse Spark Open，主打 1M 上下文和 Agent 任务，作者还算清了要跑起来得 700GB 显存，看看你的卡够不够。
Chinese contentMiMo V2.6 Pro 登顶 Agent Arena 开源模型榜首
小米 MiMo V2.6 Pro 超过 DeepSeek 登上 Agent Arena 开源第一，单任务才 $0.11，做智能体的可以看看这份榜单。
Chinese contentMicrosoft MAI-Base-2 预训练完成，FLOPs 效率达 DeepSeek-V4-Pro-Base 的 4-12 倍
微软自家基座模型 MAI-Base-2 训完了，号称算力效率是 DeepSeek-V4-Pro-Base 的 4-12 倍，还顺手把省下的算力拿去训更大的模型。
Chinese contentMicrosoft AI 称以 1/12 训练算力在代码任务上追平 DeepSeek-V4-Pro-Base
微软 AI 的 Mustafa Suleyman 说他们用少 12 倍的算力就在代码评测上追平了 DeepSeek-V4-Pro-Base，不过细节还没发，先看看数据再说。
Nemotron 3.5 Super VL Powers Deep Agents via ChatNVIDIA
Speed, not raw accuracy, is the headline here: a 55% reduction in total task time against DeepSeek V4.1 Flash suggests latency-sensitive agent pipelines are where this model fits best. Developers building Deep Agents workflows in Python can test it directly through ChatNVIDIA with open weights, making self-hosting or cost comparison against closed APIs straightforward.
10月15日
Chinese contentDeepSeek 服务恢复，疑似新模型上线，生成速度约 180 t/s
有人实测 DeepSeek 服务恢复后的新模型，速度 180 t/s，表现接近 Claude，可能是 4.1 Pro，想去试的可以看看。
DeepSeek V4.1 Flash Tops Artificial Analysis AgentPerf Leaderboard
Serving efficiency, not raw scale, is where V4.1 Flash wins: it delivers 1.3x to 6.2x the throughput per MW of V4 Pro 0813 across Blackwell hardware while scoring higher on intelligence. Teams running agentic coding workloads can use the AA-AgentPerf data to pick hardware — B300 nodes stay within about 20% of rack-scale GB300 NVL72, so single 8-GPU nodes are viable for this model. As an open-weights release from DeepSeek, it widens cheap self-hosting options for Southeast Asian inference providers competing with rented frontier APIs.
AMD MI355X joins AA-AgentPerf benchmark against NVIDIA Blackwell
The MI355X's 288GB of HBM3E per GPU gives it an edge on very large models like the 1.6T-parameter DeepSeek V4 Pro, but serving-stack maturity on ROCm remains the bottleneck at higher loads. Teams evaluating accelerator purchases for agentic inference can now compare AMD and NVIDIA on realistic coding-agent workloads rather than synthetic throughput numbers. All results run on ROCm nightly builds of vLLM and SGLang, so scores should improve as the software stack matures.
10月14日
Chinese contentDeepSeek V4.1 自评得分远高于同为 V4.1 的子智能体
有人发现 DeepSeek V4.1 给自己打分挺高，给同样是自己扮演的子智能体打分却低不少，这个自评偏差挺有意思。
Fish Audio Drama 3 Preview Tops Open-Weights TTS Leaderboard
An open-weights model beating closed leaders like Eleven v4 on pronunciation while costing a fifth of the price signals real pressure on commercial TTS pricing. Builders producing audiobooks, dubbed content or multi-speaker agents can self-host the weights from Hugging Face or call the API, with 83-language support covering major Southeast Asian languages. The Elo gain of 101 points over Fish Audio's own S2.1 Pro shows the jump is internal as well as competitive.
Chinese contentDeepSeek V4.1 Flash 在 ARC-AGI-3 得分 3.3%，评估成本 $878
ARC Prize 实测了 DeepSeek V4.1 Flash，ARC-AGI-3 跑出 3.3%，成本不到 1000 美元，目前性价比最好的一份成绩单。
10月13日
Chinese contentCodeVideo Model 项目发布：代码演示视频生成模型开源
一个叫 CodeVideo Model 的开源项目，做的是自动生成代码演示视频，项目页和教程都放出来了，做开发教程内容的可以看看。
Chinese contentV4.1 与 Sol 6.1 表现出色，Haiku 依旧不在小模型赛道
V4.1 和 Sol 6.1 这轮表现都很能打，尤其 Sol 6.1 出乎意料，还顺带聊了 Haiku 和 Fable 5.5 的定位。
FrontierCode 1.1 Leaderboard Launches on Artificial Analysis
The benchmark penalises behaviours conventional evals ignore, such as oversized diffs, out-of-scope file edits and unhelpful tests, so its ranking offers a different picture of coding ability than SWE-style test suites. Even the best model fails over 40% of tasks, meaning maintainer cleanup is still unavoidable. Teams choosing models by cost-performance should note GPT-6.1 Sol matching Claude Fable 5.1 at under a tenth of the per-task price.
Artificial Analysis launches leaderboard for Cognition's FrontierCode 1.1
The benchmark's merge-focused grading penalises the messy diffs and unhelpful tests that cheaper evaluations ignore, so leaderboard gaps here reflect real maintainer acceptance rather than raw test-passing. Leading models still fail over 40% of tasks, and nine tasks went unsolved by every model evaluated. Teams choosing coding agents can weigh GPT-6.1 Sol's 49% score at US$0.81 per task against Claude Opus 5.5's 52% at far higher cost, with open-weights options like MiMo-V2.6-Pro at 45% close behind.
Chinese contentAgent Arena 数据：开源模型 12 周追平 Claude Fable 5
LMArena 给出实测数据：Kimi K3 和 DeepSeek V4.1 Flash 用不到一成的成本追平了 Claude Fable 5，做 agent 选型前可以看看这份榜单。
Chinese contentMeta 发布 Muse Wonder 与 Muse Wonder Full，智能指数升至 52 和 56
Meta 六周内又发了新模型，智能指数涨到 56，代理类任务进步明显，比 Claude 便宜但比 GPT-6 贵，做 agent 的可以看看。
10月12日
Adaptive DHiSparse Speculative Decoding
This approach solves a critical bottleneck in long-context AI serving by enabling more efficient memory usage while maintaining high throughput. The technology allows models to handle significantly larger contexts without hitting memory limits, which directly impacts real-world applications requiring extensive context windows. Developers working with large language models should consider this optimization technique to improve performance in production environments.
10月10日
HeyGen Voice Tops Artificial Analysis Controlled Voice TTS Leaderboard
Leaderboard rank alone overstates the win: the cheaper Alibaba Qwen model actually beats HeyGen on pronunciation robustness, and Eleven v4 is nearly twice as fast, so the top spot rests on voice quality in assistant-style use cases. Teams building voice agents in US and UK English can use the Elo splits by category, plus the price and speed data, to pick a TTS vendor. Alibaba's strong showing also signals competitive Chinese TTS at lower cost.
10月9日
StoreBench Live-Commerce Agent Evaluation
StoreBench introduces a dynamic, production-grade environment that reveals significant gaps between current LLM capabilities and human performance in e-commerce operations. The benchmark's realistic challenges - including supplier failures and market shocks - provide a more rigorous test than static evaluations. Researchers and developers can access five training tasks and sample trajectories to improve agent performance in complex, time-sensitive commercial scenarios.
DeepSeek V4.1 Flash n-gram editing
This approach offers a lightweight way to update models without full retraining, potentially reducing costs for personalized AI applications. Developers working with large language models could implement this technique for efficient knowledge updates, especially useful for applications requiring frequent factual updates or user-specific customization.
10月8日
Anthropic's Haiku 5.5 undercuts DeepSeek and GLM on price
The takeaway is that Anthropic is now competing on price, not just capability, in the small-model tier where Chinese labs have dominated. Builders choosing a cheap workhorse model for high-volume tasks can add Haiku 5.5 to a shortlist alongside DeepSeek-V4.1 flash and GLM 5.3 flash and re-run their own workloads to compare latency and cost. Pricing and availability for ASEAN API users should be checked against local provider rates.
vLLM Highlights DeepSeek-V4.1-Flash SWA Optimization
SWA bounded replay shows how KV-cache reuse for agent loops can be engineered around sliding-window attention, a pattern many teams will copy as agent traffic grows. Engineers running agent stacks on vLLM can apply it to reduce TTFT under high concurrency. The writeup also signals that DeepSeek-V4.1-Flash inference performance is being actively tuned by the community rather than relying on the lab alone.
DeepSeek cuts agent memory cost with new Flash model
Sparse agent memory changes the economics of long-context agent runs, since cache footprint rather than raw context length becomes the cost driver. Developers running multi-step agent pipelines can weigh Flash's cheaper, lighter memory against the extra per-token compute it needs. At US$0.27 versus US$0.67, it also undercuts DeepSeek's own V4-Pro on price while scoring higher on the AA Index.
