全部 AI 动态

Sakana Fugu surprisingly performed near GLM 5.2 level but 17× more expensive! We gave the same prompt to 4 models: build a complete live Trader Desk with both frontend and backend components, real-time market data fetched from external APIs for 8 symbols, and a custom dark-theme UI. Outputs: Fugu Ultra — 22,225 t, $0.51 Opus 4.8 — 15,802 t, $0.31 GPT-5.5 — 11,474 t, $0.26 GLM 5.2 — 13,677 t, $0.03 Fugu created the most polished and feature-rich trading desk in the run. GLM 5.2 was very close behind, with a similarly complete multi-panel interface and live data, but at a much lower cost. Opus and GPT also performed well, delivering solid results with a better balance between quality and cost

http://x.com/i/article/2067849694232080384

many people asked me to make a video about my complete agentic engineering workflow excited to share it's finally here!!! it took me about 20 hours in total to record this 45 minutes of walkthrough - it covers everything i do to ship production quality code at an average 40+ PRs/day velocity hope this can be a useful reference to everyone exploring good ways to use AI. and would appreciate a reshare with anyone you think might benefit from this! enjoy! https://youtu.be/iQyg-KypKAA

这个 Codex 里的“循环”自动化简直太疯狂了。 “/goal 逐一检查这个应用里的每一个功能，根据代码创建带有预期行为的用户故事，维护一张单一的规范电子表格来跟踪功能状态 - 完成后，将循环切换为测试每一个用户故事并记录所有错误 - 完成后，修复每一个逻辑错误或 UX 错误 - 修复后再次测试每一个用户行为” 感谢 @MatthewBerman 的提醒。 数百个用户故事被轻松处理，就像不费吹灰之力。

Built a vision-agent security monitoring demo with MiniCPM-V 4.6. @OpenBMB Four CCTV style feeds run on a live clock. MiniCPM watches recent frame windows, decides whether activity is normal, and when it sees a genuine incident, it streams a tool call: raise_alert(reason, severity) The goal was to test whether a small VLM can move beyond captioning into practical agent behavior: observe pixels, reason about context, suppress normal activity, and trigger a structured action only when needed. What stood out: MiniCPM-V 4.6 handled the workflow well enough to feel useful for real world monitoring prototypes. Not just “what is in the image?”, but “should anything be done?” Small VLMs are getting very interesting for edge-ish and operational workflows.

Built a back-office agent swarm demo with MiniCPM5-1B. 128 concurrent agents on DGX Spark, served through vLLM continuous batching. Each agent handled a different business ops queue: invoice review refund routing compliance checks HR/payroll issues legal triage IT security alerts sales ops logistics recovery MiniCPM5-1B streamed 6,604 chunks across 128 agents in 1.48s. The takeaway: The value of 1B models is not just “can it answer well?” It is: “How many useful business decisions can it make at once?” Not one giant agent doing everything slowly. A swarm of tiny, cheap workers clearing practical queues in parallel. cc: @OpenBMB @NVIDIAAI

A gift to the Chinese developer community for anyone building agents on @LangChain: After 2 books on the ecosystem, I open-sourced the third — Deep Agents in Action. 8 chapters live and counting, on the Agent Harness: planning, context engineering, sub-agents, Skills, memory.

http://vercel.com/design.md

Pixverse’s preset effects are seriously cool 🔥 Example with the Dark Pact preset: you just upload your photo… and BOOM, you get a stylish video instantly. Watch this 👇 Super simple and the result is clean. Have you tried Pixverse presets yet?

been testing MiniCPM-V 4.6 locally on my DGX Spark this demo is an industrial gauge reader i generated a synthetic control panel with a pressure gauge, temperature gauge, flow display, and tank level bar. because the panel is synthetic, i know the exact ground truth at every frame. MiniCPM gets a frame and has to return structured JSON: pressure_bar temp_c flow_lpm level_pct then i score each reading live: pass = within 5% of full scale drift = within 10% miss = outside that the scenario i was targeting is industrial visual ops. lots of real systems still have gauges, panels, status screens, meters, and legacy interfaces. i wanted to see if a small open source vision model could read that kind of interface and return values another system could actually use. what i saw in practice: digital displays and tank level were easier. analog gauges were harder, especially between tick marks or while moving. but the important part is that the output is measurable. it is not just a captioning demo. MiniCPM-V 4.6 had to read pixels, follow a schema, and produce structured data i could check against ground truth.

http://x.com/i/article/2067278912984436736

Frontier doesn't only mean Anthropic and OpenAI anymore. I built this black hole simulator by simply dropping an illustration screenshot into Kilo Code, switching to @MiniMax_AI M3, and prompting it to "animate this screenshot into a working black hole simulator". M3's vision capabilities are incredible. Total cost: $0.53 Speaking of stars - Kilo Code is about to hit 25k, and we're giving $500 of AI Credits to 2 people once that happens! Check out our repo here: https://github.com/Kilo-Org/kilocode/

Frontier doesn't only mean Anthropic and OpenAI anymore. I built this black hole simulator by simply dropping an illustration screenshot into Kilo Code, switching to @MiniMax_AI M3, and prompting it to "animate this screenshot into a working black hole simulator". M3's vision capabilities are incredible. Total cost: $0.53 Speaking of stars - Kilo Code is about to hit 25k, and we're giving $500 of AI Credits to 2 people once that happens! Check out our repo here: https://github.com/Kilo-Org/kilocode/

Love this direction from OpenAI. That’s what we call being open. In practice, though, many third-party APIs are still Chat Completions-compatible rather than Responses-compatible, which makes direct usage with Codex tricky. CC Switch tries to solve exactly this gap and make more models usable in Codex. 100% open-source and free, give it a try!