# ![@SkylerMiao7 Avatar](https://lunarcrush.com/gi/w:26/cr:twitter::1879314108723658752.png) @SkylerMiao7 Skyler Miao

Skyler Miao posts on X about agents, in the, $2413t, ai the most. They currently have [-----] followers and [---] posts still getting attention that total [-------] engagements in the last [--] hours.

### Engagements: [-------] [#](/creator/twitter::1879314108723658752/interactions)
![Engagements Line Chart](https://lunarcrush.com/gi/w:600/cr:twitter::1879314108723658752/c:line/m:interactions.svg)

- [--] Week [-------] +205%
- [--] Month [-------] -25%

### Mentions: [--] [#](/creator/twitter::1879314108723658752/posts_active)
![Mentions Line Chart](https://lunarcrush.com/gi/w:600/cr:twitter::1879314108723658752/c:line/m:posts_active.svg)

- [--] Month [--] -80%

### Followers: [-----] [#](/creator/twitter::1879314108723658752/followers)
![Followers Line Chart](https://lunarcrush.com/gi/w:600/cr:twitter::1879314108723658752/c:line/m:followers.svg)

- [--] Week [-----] +6%
- [--] Month [-----] +19%

### CreatorRank: [------] [#](/creator/twitter::1879314108723658752/influencer_rank)
![CreatorRank Line Chart](https://lunarcrush.com/gi/w:600/cr:twitter::1879314108723658752/c:line/m:influencer_rank.svg)

### Social Influence

**Social category influence**
[technology brands](/list/technology-brands)  4% [finance](/list/finance)  4% [social networks](/list/social-networks)  3% [travel destinations](/list/travel-destinations)  2% [gaming](/list/gaming)  1% [cryptocurrencies](/list/cryptocurrencies)  1% [currencies](/list/currencies)  1% [exchanges](/list/exchanges)  1% [nfts](/list/nfts)  1%

**Social topic influence**
[agents](/topic/agents) #105, [in the](/topic/in-the) #3435, [$2413t](/topic/$2413t) 6%, [ai](/topic/ai) 5%, [core](/topic/core) 5%, [vibe coding](/topic/vibe-coding) 4%, [strong](/topic/strong) 4%, [inference](/topic/inference) 4%, [agentic](/topic/agentic) 3%, [llm](/topic/llm) 3%

**Top accounts mentioned or mentioned by**
[@minimaxai](/creator/undefined) [@steipete](/creator/undefined) [@kilocode](/creator/undefined) [@openrouterai](/creator/undefined) [@roocode](/creator/undefined) [@cline](/creator/undefined) [@opencode](/creator/undefined) [@lohigowdain](/creator/undefined) [@moltbot](/creator/undefined) [@yongyi_xu](/creator/undefined) [@erikwangx](/creator/undefined) [@prepuxyo](/creator/undefined) [@zpysky1125](/creator/undefined) [@ftlognotmark](/creator/undefined) [@jenzhuscott](/creator/undefined) [@minimaxagent](/creator/undefined) [@ivanfioravanti](/creator/undefined) [@cloudxdev](/creator/undefined) [@davidszd1](/creator/undefined) [@brandgrowthos](/creator/undefined)

**Top assets mentioned**
[The 99% Community (99)](/topic/$99)
### Top Social Posts
Top posts by engagements in the last [--] hours

"grab minimax desktop now and be ready when it hits 🚀 https://agent.minimax.io/download https://agent.minimax.io/download"  
[X Link](https://x.com/SkylerMiao7/status/2021412228101243335)  2026-02-11T02:33Z [----] followers, [---] engagements


"A glance of MiniMax [---] are you ready"  
[X Link](https://x.com/SkylerMiao7/status/2021578926884053084)  2026-02-11T13:35Z [----] followers, 156.6K engagements


"Honestly I wanna release this beast ASAP Im dying to go back to my hometown for Spring Festival 😂 But the more training compute we put in the more it keeps rising. Painfully happy problem. We hear you guys. M2.5 soon. A glance of MiniMax [---] are you ready https://t.co/kZ9UCpz2Yt A glance of MiniMax [---] are you ready https://t.co/kZ9UCpz2Yt"  
[X Link](https://x.com/SkylerMiao7/status/2021587213230715306)  2026-02-11T14:08Z [----] followers, 72K engagements


"openclaw MiniMax M2.1 free 😎 7-day Coding Plan exclusive for clawdbot users. spent a lot of time with @steipete to make it. one command. one-click login. no complex setup. youre in. enjoy your claw 🦞 curl -fsSL bash https://github.com/openclaw/openclaw/releases/tag/v2026.1.30 https://skyler-agent.github.io/oclaw/i.sh https://github.com/openclaw/openclaw/releases/tag/v2026.1.30 https://skyler-agent.github.io/oclaw/i.sh"  
[X Link](https://x.com/SkylerMiao7/status/2017789329138212986)  2026-02-01T02:37Z [----] followers, 76K engagements


"yep it began as an experiment. m2.5 is shaping up to be even crazier. well probably run another AMA then😉 @prepuxyo The sparsity of this model is mind-blowing given how smart and capable it is. Can you believe that this was a side-project for @MiniMax_AI CC: @SkylerMiao7 @zpysky1125 https://t.co/L9ihhnr9Wv https://t.co/VAbynZcumE @prepuxyo The sparsity of this model is mind-blowing given how smart and capable it is. Can you believe that this was a side-project for @MiniMax_AI CC: @SkylerMiao7 @zpysky1125 https://t.co/L9ihhnr9Wv https://t.co/VAbynZcumE"  
[X Link](https://x.com/SkylerMiao7/status/2018130019722367299)  2026-02-02T01:11Z [----] followers, [----] engagements


"@ftlog_not_mark lol actually testing a mid-checkpoint rn and it's surprisingly good. more soon. stay tuned 👀"  
[X Link](https://x.com/SkylerMiao7/status/2019700837677953451)  2026-02-06T09:12Z [----] followers, [--] engagements


"Honestly I wanna release this beast ASAP Im dying to go back to my hometown for Spring Festival 😂 But the more training compute we put in the more it keeps rising. Painfully happy problem. We hear you guys. M2.5 soon"  
[X Link](https://x.com/SkylerMiao7/status/2021587162248966526)  2026-02-11T14:08Z [----] followers, [----] engagements


"Happy Chinese New Year Highly anticipated opensource models dropping this week DeepSeek-V4 GLM-5 MiniMax-M2.5 Qwen-3.5 Highly anticipated opensource models dropping this week DeepSeek-V4 GLM-5 MiniMax-M2.5 Qwen-3.5"  
[X Link](https://x.com/SkylerMiao7/status/2021605643702902889)  2026-02-11T15:21Z [----] followers, 11.5K engagements


"Enjoy @SkylerMiao7 @steipete Thanks Man till [--] day i can enjoy 🫡 https://t.co/A5iD5j8die @SkylerMiao7 @steipete Thanks Man till [--] day i can enjoy 🫡 https://t.co/A5iD5j8die"  
[X Link](https://x.com/anyuser/status/2018130512964174270)  2026-02-02T01:13Z [----] followers, [----] engagements


"🚨 Accounts got hacked again. Any crypto-related posts or messages from me and @MiniMax_AI are fake. Please ignore. And people wonder why I don't support crypto. Got so much love from that community. "Lobster from the base" could be a really cool band tho. https://t.co/7pOa0pdmG4 And people wonder why I don't support crypto. Got so much love from that community. "Lobster from the base" could be a really cool band tho. https://t.co/7pOa0pdmG4"  
[X Link](https://x.com/SkylerMiao7/status/2019235744234238174)  2026-02-05T02:24Z [----] followers, [----] engagements


"Huge congrats to @kilocode @MiniMax_AI M2.1 is now live as the default model on Kilo CLI Give it a spin and feel the difference🦾 Kilo CLI [---] brings open source vibe coding to your terminal with support for 500+ models https://t.co/8Lw0wyCdnU Kilo CLI [---] brings open source vibe coding to your terminal with support for 500+ models https://t.co/8Lw0wyCdnU"  
[X Link](https://x.com/SkylerMiao7/status/2019243982757437848)  2026-02-05T02:57Z [----] followers, 17K engagements


"DEFAULT and FREE on @kilocode now Reward yourself with a quick game of space invaders after a long session of watching M2.1 code. DEFAULT and FREE on @kilocode now https://t.co/zQDRRo93N6 Reward yourself with a quick game of space invaders after a long session of watching M2.1 code. DEFAULT and FREE on @kilocode now https://t.co/zQDRRo93N6"  
[X Link](https://x.com/SkylerMiao7/status/2019251580173008929)  2026-02-05T03:27Z [----] followers, [----] engagements


"Intelligence with Everyone First the good part of the Anthropic ads: they are funny and I laughed. But I wonder why Anthropic would go for something so clearly dishonest. Our most important principle for ads says that we wont do exactly this; we would obviously never run ads in the way Anthropic First the good part of the Anthropic ads: they are funny and I laughed. But I wonder why Anthropic would go for something so clearly dishonest. Our most important principle for ads says that we wont do exactly this; we would obviously never run ads in the way Anthropic"  
[X Link](https://x.com/SkylerMiao7/status/2019364875009683509)  2026-02-05T10:57Z [----] followers, 62K engagements


"Get started in seconds with MiniMax OAuth. No API keys to copy https://t.co/K9YAoLzEKA https://t.co/K9YAoLzEKA"  
[X Link](https://x.com/SkylerMiao7/status/2019383700665037204)  2026-02-05T12:12Z [----] followers, [----] engagements


"I like Tokyo more One-shot one city. https://t.co/dNVmN9afb2 https://t.co/svgsycIEcZ One-shot one city. https://t.co/dNVmN9afb2 https://t.co/svgsycIEcZ"  
[X Link](https://x.com/SkylerMiao7/status/2019792009863807003)  2026-02-06T15:15Z [----] followers, [----] engagements


"RT @MiniMax_AI: Kevin Parker just felt correct for this universe"  
[X Link](https://x.com/SkylerMiao7/status/2019934653105492249)  2026-02-07T00:42Z [----] followers, [--] engagements


"RT @jenzhuscott: Again: [--]. @MiniMax_AI released this. Stop lumping everything under a nation. You don't call "US released xyz" whenever Goo"  
[X Link](https://x.com/SkylerMiao7/status/2021141824497385900)  2026-02-10T08:38Z [----] followers, [--] engagements


"cowork just shipped windows support good for them. meanwhile minimax desktop users been vibing on windows for a couple weeks already 😎 and with M2.5 dropping soon you ain't seen nothing yet welcome new player to windows world 🪟 MiniMax Agent = Claude cowork + Clawdbot - Lives on your computer AND in the cloud - Agentic + batch workflows - Full browser control - Agent skills library - Expert agents: customized & shared - Runs on macOS & Windows Upgrade your tools and mindset. Everyone is an Agent https://t.co/odFax7NY5v MiniMax Agent = Claude cowork + Clawdbot - Lives on your computer AND in"  
[X Link](https://x.com/SkylerMiao7/status/2021412225832059367)  2026-02-11T02:33Z [----] followers, [----] engagements


"@MiniMaxAgent"  
[X Link](https://x.com/SkylerMiao7/status/2021417846572712001)  2026-02-11T02:55Z [----] followers, [---] engagements


"RT @ivanfioravanti: In the past days I've got early access to MiniMax M2.5 and I've been able to play with it quite a lot. M2.1 was already"  
[X Link](https://x.com/anyuser/status/2021606490620985364)  2026-02-11T15:25Z [----] followers, [--] engagements


"you guys are probably tired of me yapping about code and agents. but MiniMax-M2.5 isn't just SOTA at coding it's a legit workspace companion. Research organize deliver. All from one prompt. coming soon. MiniMax Agent just turned one prompt into a full recruiting playbook. 📖 [--] schools. Career fair schedules. Costs. Registration links. All-in-one Excel researched and organized in minutes. No Googling no spreadsheet wrangling. Just ask and it's done. Try it today https://t.co/BzPMsShzAy MiniMax Agent just turned one prompt into a full recruiting playbook. 📖 [--] schools. Career fair schedules."  
[X Link](https://x.com/SkylerMiao7/status/2021936738793636306)  2026-02-12T13:17Z [----] followers, [----] engagements


"RT @cloudxdev: Turning a book into Notes Mind Maps Slides with @MiniMax_AI Desktop Agent I have attached a book pdf (320 pages) into Mi"  
[X Link](https://x.com/SkylerMiao7/status/2021940183466488318)  2026-02-12T13:31Z [----] followers, [--] engagements


"Happy Thanksgiving Super grateful and excited to play even a small role in something game-changing and to keep pushing technological innovation forward🤲"  
[X Link](https://x.com/SkylerMiao7/status/1994236804137144385)  2025-11-28T02:47Z [----] followers, [----] engagements


"@DavidSZD1 Will be in M3"  
[X Link](https://x.com/SkylerMiao7/status/1995337260208955520)  2025-12-01T03:40Z [----] followers, [--] engagements


"Great to see DeepSeek V3.2 supporting interleaved thinking MiniMax has been working with the community @kilocode @roocode @cline @OpenRouterAI and many more to build solid unified support for interleaved thinking. Now all users and new interleaved-thinking models can benefit from a smoother consistent experience across tools. 🚀 💻 API Update 🔹 V3.2: Same usage pattern as V3.2-Exp. 🔹 V3.2-Speciale: Served via a temporary endpoint: base_url="https://t.co/WoEd0WDQ5x". Same pricing as V3.2 no tool calls available until Dec 15th [----] 15:59 (UTC Time). 💡 V3.2 now supports Thinking in Tool-Use"  
[X Link](https://x.com/SkylerMiao7/status/1995457917567533295)  2025-12-01T11:40Z [----] followers, 146.4K engagements


"@BrandGrowthOS @OpenRouterAI @kilocode @cline @roocode @vercel Agreed. Interleaved thinking isnt just about better outputs it keeps agents from drifting mid-run which cuts down human fixes and review loops before shipping"  
[X Link](https://x.com/SkylerMiao7/status/2001193064128237733)  2025-12-17T07:29Z [----] followers, [--] engagements


"Ive been using M2.1 (latest ckpt for internal test) full-time over the past few days purely as a developer not training it just writing and shipping code with it. Its not perfect yet but the jump from M2 M2.1 is very noticeable. Im now comfortable relying on it for real engineering work: [--] vibe coding (not only web but also mobile apps) Feature development refactors bug fixes and code review in large existing repositories One concrete example from my own project: before refactoring I asked M2.1 GPT-5.2 Opus [---] (High & Low) and Gemini [--] Pro to review a 1300-line Python file (shell_tool.py)"  
[X Link](https://x.com/SkylerMiao7/status/2001241040532201827)  2025-12-17T10:40Z [----] followers, 35.5K engagements


"Solid take. MiniMax-M2 is optimized for agentic workflows coding and long-horizon tasks with interleaved thinking. Nice to see it performing well in home-lab setups thanks to the local LLM crowd for pushing it in real workflows. top [--] local LLMs to run at home for December [----] - updated Devstral-2-123B at 1st place senior engineer energy not flashy just relentlessly effective this is the model i trust when i need results just fix the thing agentic coding is the real superpower clean https://t.co/dlkQHy2vev top [--] local LLMs to run at home for December [----] - updated Devstral-2-123B at 1st"  
[X Link](https://x.com/SkylerMiao7/status/2001551440159047813)  2025-12-18T07:13Z [----] followers, [----] engagements


"@koylanai Cant wait to see what you build with M2.1 🚀 In my test it already feels better with Cursor 1.x. Cursor 2.x changed quite a bit on their side Im still testing and will share updates once I have a clearer signal"  
[X Link](https://x.com/SkylerMiao7/status/2001559140502839297)  2025-12-18T07:44Z [----] followers, [--] engagements


"@communicating The Prompt is like "Review the shell_tool.py tell me what problem you see and give me refactor suggestion""  
[X Link](https://x.com/SkylerMiao7/status/2001575294575239197)  2025-12-18T08:48Z [----] followers, [--] engagements


"If all goes well M2.1 early access should be available within the next [--] hours. Heres one of my favorite vibe pieces from recent tests ✨"  
[X Link](https://x.com/SkylerMiao7/status/2001875728410034497)  2025-12-19T04:42Z [----] followers, 73.1K engagements


"one-shot by M2.1 Internal testing for M2.1 👀 https://t.co/rp2vWZ49A6 Internal testing for M2.1 👀 https://t.co/rp2vWZ49A6"  
[X Link](https://x.com/SkylerMiao7/status/2001912151737614343)  2025-12-19T07:07Z [----] followers, [----] engagements


"M2.1 early access start Apply here: https://docs.google.com/forms/d/e/1FAIpQLSewx7znj2DdlQvbj6uP2lcXtJ0MQ7HzJuqNjoSrIvzgaSApUQ/viewformpli=1 If all goes well M2.1 early access should be available within the next [--] hours. Heres one of my favorite vibe pieces from recent tests ✨ https://t.co/ur5Aplnnmp https://docs.google.com/forms/d/e/1FAIpQLSewx7znj2DdlQvbj6uP2lcXtJ0MQ7HzJuqNjoSrIvzgaSApUQ/viewformpli=1 If all goes well M2.1 early access should be available within the next [--] hours. Heres one of my favorite vibe pieces from recent tests ✨ https://t.co/ur5Aplnnmp"  
[X Link](https://x.com/SkylerMiao7/status/2001976579971539235)  2025-12-19T11:23Z [----] followers, [----] engagements


"What we built M2.1 for [--] Real hard-core engineering reliable and usable in production [--] Vibe coding focusing on ideas flow and outcomes while M2.1 handles the implementation details To get there we focused on: [--]. Multi-language coding Not just Python. M2.1 shows strong gains across Rust / Java / Go / C++ / Kotlin / Obj-C / TS / JS closer to how real projects are built. https://twitter.com/i/web/status/2002303431248392645 https://twitter.com/i/web/status/2002303431248392645"  
[X Link](https://x.com/SkylerMiao7/status/2002303431248392645)  2025-12-20T09:01Z [----] followers, 22.8K engagements


"2. Web Dev & App Dev Strong at vibe coding more approachable for users who arent deep into code. Many models ignore mobile. M2.1 is significantly better on Android & iOS. Not only vibe web vibe mobile too (still more room to grow vs. web)"  
[X Link](https://x.com/SkylerMiao7/status/2002303673465188423)  2025-12-20T09:02Z [----] followers, [----] engagements


"This is awesome 🔥 M2.1 + Coding Plan really shines in these creative + technical mashups v1 already looks solid curious to see how far you push it. Keep us posted 👀 MiniMax Coding. Plan + M2.1 + Claude Code + Canaletto. An this is just v1. I had an idea to push this model to the limit stay tuned https://t.co/4BTrIVeaBf MiniMax Coding. Plan + M2.1 + Claude Code + Canaletto. An this is just v1. I had an idea to push this model to the limit stay tuned https://t.co/4BTrIVeaBf"  
[X Link](https://x.com/SkylerMiao7/status/2002306049030934830)  2025-12-20T09:12Z [----] followers, [----] engagements


"Great catch 👍 M2.1 isnt a small bump we spent a lot of time improving design sense structure and visual consistency especially for frontend work. Glad the difference shows MiniMax-M2.1 vs. MiniMax-M2: It seems MiniMax-M2.1 is not a minor update. The frontend design improvements are clear https://t.co/k951I0l4wz MiniMax-M2.1 vs. MiniMax-M2: It seems MiniMax-M2.1 is not a minor update. The frontend design improvements are clear https://t.co/k951I0l4wz"  
[X Link](https://x.com/SkylerMiao7/status/2002306342997185022)  2025-12-20T09:13Z [----] followers, [----] engagements


"Fly mode on real voxel scale no shortcuts. M2.1 is doing serious work here. Appreciate you sharing the test. Beta testing MiniMax-M2.1. Minecraft clone one-shot. Fly mode right from the start. Most AI-built Minecraft clones don't. Voxel block count crossed a threshold I've never seen before. It's not lazy. I'm not ready for M2.1 https://t.co/akZ4NOOB1z Beta testing MiniMax-M2.1. Minecraft clone one-shot. Fly mode right from the start. Most AI-built Minecraft clones don't. Voxel block count crossed a threshold I've never seen before. It's not lazy. I'm not ready for M2.1 https://t.co/akZ4NOOB1z"  
[X Link](https://x.com/SkylerMiao7/status/2002375508915343485)  2025-12-20T13:48Z [----] followers, [----] engagements


"@mweinbach How do you like it"  
[X Link](https://x.com/SkylerMiao7/status/2002515096971690495)  2025-12-20T23:02Z [----] followers, [---] engagements


""Generate an SVG of a pelican riding a bicycle the output should be a svg file" One-shot by MiniMax Agent + M2.1 🐦🚲 It even hosted a shareable HTML for me. Try it: M2.1 on MiniMax Agent is hopefully live Monday have fun https://yvwrtv5drk9t.space.minimax.io/ https://yvwrtv5drk9t.space.minimax.io/"  
[X Link](https://x.com/SkylerMiao7/status/2002583915207537123)  2025-12-21T03:36Z [----] followers, [----] engagements


"Design and visual quality were a big focus for M2.1. M2.5 will push this even further 👀 M2.1 @MiniMax__AI definitely got much better in design. Works really fast. Great work @SkylerMiao7 https://t.co/xME7nVZItX M2.1 @MiniMax__AI definitely got much better in design. Works really fast. Great work @SkylerMiao7 https://t.co/xME7nVZItX"  
[X Link](https://x.com/SkylerMiao7/status/2002588287714931153)  2025-12-21T03:53Z [----] followers, 10.2K engagements


"Notion Lite one-shot by M2.1 on MiniMax Agent ✨ Fully playable: https://kfk9nah5yri7.space.minimax.io/ https://kfk9nah5yri7.space.minimax.io/"  
[X Link](https://x.com/SkylerMiao7/status/2003035953221468588)  2025-12-22T09:32Z [----] followers, 15K engagements


"Same prompt. M2.1 just gets it done. Thanks for the side-by-side test 🙏 @MiniMax__AI M2.1 VS Gemini [--] Pro Same Prompt M2.1 generated an HTML file and open it using just [--] sentence [--] Pro + [---] Flash required [--] sentences didnt open it auto and when I opened it via localhost there was nothing display so I had to tell it to open it directly as HTML https://t.co/onKpktuYUO @MiniMax__AI M2.1 VS Gemini [--] Pro Same Prompt M2.1 generated an HTML file and open it using just [--] sentence [--] Pro + [---] Flash required [--] sentences didnt open it auto and when I opened it via localhost there was nothing display"  
[X Link](https://x.com/SkylerMiao7/status/2003054870006554626)  2025-12-22T10:47Z [----] followers, [----] engagements


"Vibe your own masterpiece with M2.1 🎨 Built and played with on MiniMax Agent: Check out these creations 👇 http://agent.minimax.io http://agent.minimax.io"  
[X Link](https://x.com/SkylerMiao7/status/2003127508280242554)  2025-12-22T15:36Z [----] followers, 47.4K engagements


"MiniMax M2.1 unlocked. Go build something bold beautiful and fun. https://agent.minimax.io We're cooked. [--] wild web UI examples generated by MiniMax-M2.1. The design logic is getting too good. [--]. EV web design: 3D WebGL HUD overlays and scrollytelling https://t.co/VI6YIN5mgQ https://t.co/2R9sKpLlcO https://agent.minimax.io We're cooked. [--] wild web UI examples generated by MiniMax-M2.1. The design logic is getting too good. [--]. EV web design: 3D WebGL HUD overlays and scrollytelling https://t.co/VI6YIN5mgQ https://t.co/2R9sKpLlcO"  
[X Link](https://x.com/SkylerMiao7/status/2003145956733386860)  2025-12-22T16:49Z [----] followers, [----] engagements


"MiniMax M2.1 is officially live 🚀 A SOTA OSS coding & agent model across SWE-Verified SWE-Multilingual Multi-SWE VIBE-Bench and Terminal-Bench [---]. Built for real-world coding from vibe builds to serious engineering"  
[X Link](https://x.com/SkylerMiao7/status/2003314594287354086)  2025-12-23T03:59Z [----] followers, [----] engagements


"You mean M3 Its shaping up to be something special"  
[X Link](https://x.com/SkylerMiao7/status/2003424154561941539)  2025-12-23T11:15Z [----] followers, 13.7K engagements


"Loved the stream @theo 🙌 Always fun to real-time builds and honest takes. M2.1 building a live UI app with quality on par with top closed models for just $0.1 in real time pretty wild. These models are so good that I'm going live during christmas break https://t.co/g3e4ts5UeB These models are so good that I'm going live during christmas break https://t.co/g3e4ts5UeB"  
[X Link](https://x.com/SkylerMiao7/status/2003652372908900593)  2025-12-24T02:22Z [----] followers, [----] engagements


"Were in the final stretch for open-sourcing M2.1. Aiming for Dec [--] morning 🎄 Hopefully a nice little Christmas gift 🎁 @MiniMax__AI thx for your efforts good job. what's the fate of the huggingface repo https://t.co/TFZwfmPv0v https://t.co/Nz3tlq21I5 @MiniMax__AI thx for your efforts good job. what's the fate of the huggingface repo https://t.co/TFZwfmPv0v https://t.co/Nz3tlq21I5"  
[X Link](https://x.com/SkylerMiao7/status/2003669609371795639)  2025-12-24T03:30Z [----] followers, 10.2K engagements


"You cant use OpenRouter inside Claude Code it only works with Anthropics native API. OpenRouter does support interleaved thinking you just need a tool that speaks the OR protocol. @winstondu @OpenRouterAI @SkylerMiao7 does the model on OpenRouter have interleaved thinking to get the best out of Claude Code @winstondu @OpenRouterAI @SkylerMiao7 does the model on OpenRouter have interleaved thinking to get the best out of Claude Code"  
[X Link](https://x.com/SkylerMiao7/status/2003681029060751472)  2025-12-24T04:15Z [----] followers, [----] engagements


"M2.1 can turn any website into a podcast Open sourcing soon🧑🎄 Nothing beats open-source MiniMax just dropped M2.1 and devs are calling it "Claude at 10% the cost." - 72.5% SWE-Multilingual. Beats Sonnet [---] - 88.6% VIBE-bench. Beats Gemini [--] Pro I used it to build an AI studio that turns any website into a podcast. 100% open-source. https://t.co/5WbwX6BSCU Nothing beats open-source MiniMax just dropped M2.1 and devs are calling it "Claude at 10% the cost." - 72.5% SWE-Multilingual. Beats Sonnet [---] - 88.6% VIBE-bench. Beats Gemini [--] Pro I used it to build an AI studio that turns any website"  
[X Link](https://x.com/SkylerMiao7/status/2004158247083954671)  2025-12-25T11:52Z [----] followers, [----] engagements


"Massive thanks to the vLLM team Merry Christmas🎄 Huge congrats to @MiniMaxAI_ on shipping M2.1 🥳 A massive step forward for open-source agents. vLLM provides Day-0 support for this release. 🚀 We are excited to empower the community to run this full-stack development powerhouse with maximum efficiency. Model Repo: https://t.co/TUp3LzxjS2 Huge congrats to @MiniMaxAI_ on shipping M2.1 🥳 A massive step forward for open-source agents. vLLM provides Day-0 support for this release. 🚀 We are excited to empower the community to run this full-stack development powerhouse with maximum efficiency."  
[X Link](https://x.com/SkylerMiao7/status/2004546780541899059)  2025-12-26T13:36Z [----] followers, [----] engagements


"This is insane 🤯 Guess I need an M3 Ultra as a Christmas gift to myself now. Huge thanks for the great work especially shipping this over the holidays MiniMax M2.1 in 4-bit cruises on an M3 Ultra with mlx-lm. Generated a space invaders game using [----] tokens at [----] tok/sec: https://t.co/kAZCmfsF7D MiniMax M2.1 in 4-bit cruises on an M3 Ultra with mlx-lm. Generated a space invaders game using [----] tokens at [----] tok/sec: https://t.co/kAZCmfsF7D"  
[X Link](https://x.com/SkylerMiao7/status/2004745817903743008)  2025-12-27T02:47Z [----] followers, [----] engagements


"@AonSayyed @steipete @ssh_sharma135 @Zai_org @droid @MiniMax__AI We do support count token api on Anthropic API let me check the openai API"  
[X Link](https://x.com/SkylerMiao7/status/2004752002048557129)  2025-12-27T03:11Z [----] followers, [--] engagements


"M2.1 + SGLang is a great combo for agentic workflows. Easy to serve easy to extend easy to build real things. 🎉 Congrats to @MiniMax__AI on releasing MiniMax M2.1 open source Its a strong model for real-world development and a great fit for agent workflows. Were excited to support M2.1 from day [--] in SGLang. If youre building agents using tools or chaining multi-step reasoning https://t.co/sI9NUezrcJ 🎉 Congrats to @MiniMax__AI on releasing MiniMax M2.1 open source Its a strong model for real-world development and a great fit for agent workflows. Were excited to support M2.1 from day [--] in"  
[X Link](https://x.com/SkylerMiao7/status/2004754955891331582)  2025-12-27T03:23Z [----] followers, [----] engagements


"Reason 1: Memory access math doesnt favor int4 on M2 The core benefit of int4 QAT is reducing weight memory access. For M2 deployment: [--] H200 GPUs [---] TB total VRAM Model weights: [---] GB KV cache + activations: [------] GB ➡ One inference round touches [--] TB VRAM total ➡ int4 QAT only saves [---] GB on weights So even before extra kernel / implementation overhead the upper-bound gain is 1015%. With techniques like LargeEP the benefit shrinks further"  
[X Link](https://x.com/SkylerMiao7/status/2004853974831432161)  2025-12-27T09:56Z [----] followers, [---] engagements


"Good question. In fact as early as abab5.5 (late 2023) we had already achieved near-lossless int4 PTQ and int4 QAT. Since early [----] all MiniMax models used in production inference have been deployed with: W4A8 (int4-int8) or W4A16 (int4-bf16 / fp16) Compared to PTQ our QAT generalizes extremely well we could migrate it to new models with almost no extra effort. This stayed true up until MiniMax M1. So why not int4 QAT on M2 Two main reasons @SkylerMiao7 CongratzAny chance you are looking into [--] bit QAT or reasons you preferred not to @SkylerMiao7 CongratzAny chance you are looking into [--] bit"  
[X Link](https://x.com/SkylerMiao7/status/2004854771895992558)  2025-12-27T10:00Z [----] followers, [---] engagements


"Reason 1: Memory access math doesnt favor int4 on M2 The core benefit of int4 QAT is reducing weight memory access. For M2 deployment: [--] H200 GPUs [---] TB total VRAM Model weights: [---] GB KV cache + activations: [------] GB ➡ One inference round touches [--] TB VRAM total ➡ int4 QAT only saves [---] GB on weights So even before extra kernel / implementation overhead the upper-bound gain is 1015%. With techniques like LargeEP the benefit shrinks further"  
[X Link](https://x.com/SkylerMiao7/status/2004854774467100702)  2025-12-27T10:00Z [----] followers, [---] engagements


"Reason 2: Native FP8 + kernel overhead M2 uses native FP8 training so we get W8A8 (fp8-fp8) inference for free. If we added int4 QAT the best accuracy recipe would be: 👉 W4A8 (int4-fp8) But in practice: int4-fp8 CUDA kernels have significant overhead Gains are not always faster than fp8 With short inputs inference can even become compute-bound not memory-bound We also experimented with W4A4 but havent found anything production-reliable yet"  
[X Link](https://x.com/SkylerMiao7/status/2004854776992018899)  2025-12-27T10:00Z [----] followers, [---] engagements


"Good question. As early as abab5.5 (late 2023) we achieved near-lossless int4 PTQ and int4 QAT. Our QAT generalizes well and can be migrated to new models with almost no extra effort and it remained in use through MiniMax M1. @SkylerMiao7 CongratzAny chance you are looking into [--] bit QAT or reasons you preferred not to @SkylerMiao7 CongratzAny chance you are looking into [--] bit QAT or reasons you preferred not to"  
[X Link](https://x.com/SkylerMiao7/status/2004887153290117599)  2025-12-27T12:08Z [----] followers, [----] engagements


"Yep no weekly limit. SOTA model for pro devs. Just $9.9. I finished 50% of the [--] hours limit in [---] hours running two agents parallel running similarly about how I run the $100 max plan of Claude Code For $10/month this is *too* cheap. [---] prompts seems to be an underestimate. They are counting less lol. I can do [--] prompts and https://t.co/FCjitedGcT I finished 50% of the [--] hours limit in [---] hours running two agents parallel running similarly about how I run the $100 max plan of Claude Code For $10/month this is *too* cheap. [---] prompts seems to be an underestimate. They are counting less"  
[X Link](https://x.com/SkylerMiao7/status/2004890514076909749)  2025-12-27T12:22Z [----] followers, 29K engagements


"Strongly relate. Ive been an engineer for over a decade and have always been confident in my skills. Over the past few months though Ive been going through exactly what Boris described a real paradigm shift. From Ill just do this myself to Wait an agent can probably handle this. From copiloting with AI inside an IDE to firing off [--] requests to agents in internal Slack then reviewing PRs as they come in barely opening an IDE anymore. Im building an internal agent employee system. Were now operating in an agent-centric loop where agents can evolve themselves: discuss ideas with agents agents"  
[X Link](https://x.com/SkylerMiao7/status/2005084972991000744)  2025-12-28T01:14Z [----] followers, 85K engagements


"Good question. I cant speak for other models but for MiniMax M2.1 what Artificial Analysis measures is exactly the same public model everyone uses. We dont provide AA with any special or faster version. @SkylerMiao7 Hey can you tell me why M2.1 is faster for me when i try it ( I subscribed to both GLM and Minimax). But in Artificial Analysis website it said that its slower @SkylerMiao7 Hey can you tell me why M2.1 is faster for me when i try it ( I subscribed to both GLM and Minimax). But in Artificial Analysis website it said that its slower"  
[X Link](https://x.com/SkylerMiao7/status/2005199024585122002)  2025-12-28T08:47Z [----] followers, [----] engagements


"Feel exactly the same. I still deeply love writing code by hand especially the core parts. That joy is something Ill never give up. At the same time having frontier models inside agents working asynchronously alongside me on new ideas refactors and bug fixes honestly feels like having a few extra selves. So many ideas that used to stay in my head now actually ship. It feels great. Thats the experience we built MiniMax M2.1 for and seeing it work on real codebases is incredibly rewarding. https://twitter.com/i/web/status/2005520069539885189 https://twitter.com/i/web/status/2005520069539885189"  
[X Link](https://x.com/SkylerMiao7/status/2005520069539885189)  2025-12-29T06:03Z [----] followers, [----] engagements


"Same preferences here. I still love writing core code by hand. That joy never goes away. Frontier models inside agents working asynchronously feel like having a few extra selves ideas that used to stay in my head now actually ship. Thats exactly the experience we built MiniMax M2.1 for. Opus Gemini [--] and MiniMax M2.1 are the first models I've thrown at major code bases like Rails and Basecamp where I've been genuinely impressed. By no means perfect and you couldn't just let them vibe but the speed-up is now undeniable. Opus Gemini [--] and MiniMax M2.1 are the first models I've thrown at major"  
[X Link](https://x.com/SkylerMiao7/status/2005520922627670431)  2025-12-29T06:07Z [----] followers, [---] engagements


"Same preferences here I still love writing core code by hand And frontier models inside agents working asynchronously feel like having a few extra selves ideas that used to stay in my head now actually ship Thats exactly what we built MiniMax M2.1 for Opus Gemini [--] and MiniMax M2.1 are the first models I've thrown at major code bases like Rails and Basecamp where I've been genuinely impressed. By no means perfect and you couldn't just let them vibe but the speed-up is now undeniable. Opus Gemini [--] and MiniMax M2.1 are the first models I've thrown at major code bases like Rails and Basecamp"  
[X Link](https://x.com/SkylerMiao7/status/2005523134573891772)  2025-12-29T06:15Z [----] followers, [----] engagements


"Powerful accessible affordable open source. MiniMax M2.1 is our way of putting frontier intelligence into more hands. @Kajin_0 That's why I mentioned MiniMax M2.1. It's way way cheaper than Opus say. Would give that a try @Kajin_0 That's why I mentioned MiniMax M2.1. It's way way cheaper than Opus say. Would give that a try"  
[X Link](https://x.com/SkylerMiao7/status/2005524879970611544)  2025-12-29T06:22Z [----] followers, [----] engagements


"Great to see more testing with M2.1. Actually Ruby performance is one of the areas were pushing hard in M2.x 😎 Working on ActionMCP for Rails. Started off asking Gemini [--] Codex Opus and M2.1 to give me an implementation. All of them got something working but all took very different paths. Fascinating Still going to rewrite the whole thing by hand though but the sketches helped. Working on ActionMCP for Rails. Started off asking Gemini [--] Codex Opus and M2.1 to give me an implementation. All of them got something working but all took very different paths. Fascinating Still going to rewrite the"  
[X Link](https://x.com/SkylerMiao7/status/2005974095259975749)  2025-12-30T12:07Z [----] followers, [----] engagements


"MiniMax-01 technical report made it into Hugging Faces [----] Top [--] upvoted papers. Appreciate the love from researchers and the community. M2 uses a different architecture but we learned a ton from [--] and M1 in practice and poured those lessons into the M2/M2.1 series. We wouldnt have gotten M2.1 to this level without it. It's time for @huggingface Daily Papers wrapped [----] 🎁 Here are the Top [--] most upvoted papers: - Efficient LLM Post-Training by @gensynai - the Missing Link between the Transformer and Models of the Brain - Tiny Recursive Model (TRM) by @jm_alexia - technical reports of"  
[X Link](https://x.com/SkylerMiao7/status/2006235896522359284)  2025-12-31T05:28Z [----] followers, [----] engagements


"Yep. M2.1 coding plan is honestly a no-brainer. Worth trying tested the model for a bit & it's a no brainer to get the coding plan $2 and it plugs right into claude code harness thoughts - very fast - much much better at following directions than previous release i would put it a tier under sonnet [---] but the speed makes up for it https://t.co/QcYKyLnma0 tested the model for a bit & it's a no brainer to get the coding plan $2 and it plugs right into claude code harness thoughts - very fast - much much better at following directions than previous release i would put it a tier under sonnet 4.5"  
[X Link](https://x.com/SkylerMiao7/status/2007291079507489199)  2026-01-03T03:21Z [----] followers, [----] engagements


"@yongyi_xu Thanks for feedback. Could you DM me your groupid the screenshot issue time. We will look into it"  
[X Link](https://x.com/SkylerMiao7/status/2007641636373930208)  2026-01-04T02:34Z [----] followers, [--] engagements


"We believe coding agents are still at an early stage of their journey. In [----] well keep exploring new and exciting directions. Hope this meets your needs and goals too Definitely worth a read🔥 https://t.co/PEUv5PNBmN https://t.co/PEUv5PNBmN"  
[X Link](https://x.com/SkylerMiao7/status/2007845807010283797)  2026-01-04T16:05Z [----] followers, [----] engagements


"Its a wild beast ⚡ Glad youre enjoying the speed and intelligence. Top-tier performance on just 83090s exactly the efficiency were pushing for. MiniMax is blazing fast on local hardware getting that is wild. Nothing comes close at this level of intelligence https://t.co/vf9yMbBXsJ MiniMax is blazing fast on local hardware getting that is wild. Nothing comes close at this level of intelligence https://t.co/vf9yMbBXsJ"  
[X Link](https://x.com/SkylerMiao7/status/2004538775112978754)  2025-12-26T13:04Z [----] followers, [----] engagements


"A lot of M2.1 users say opencode is their favorite tool to pair with. M2.1 + opencode is a strong combo would love to see more user results and feedback. we have both glm [---] and m2.1 free right now curious how they stack up against each other we have both glm [---] and m2.1 free right now curious how they stack up against each other"  
[X Link](https://x.com/SkylerMiao7/status/2005794346432659868)  2025-12-30T00:13Z [----] followers, 15.6K engagements


"Have fun And tell us what should we push hardest in the next M2.x ✨ Happy New Year Arena fam ✨ Thank you for building experimenting and voting with us all year long. From websites to apps to games your Code Arena creations showed the human side of frontier AI. Were kicking off [----] by highlighting the Top [--] models in Code Arena https://t.co/wSuS7NtcqG ✨ Happy New Year Arena fam ✨ Thank you for building experimenting and voting with us all year long. From websites to apps to games your Code Arena creations showed the human side of frontier AI. Were kicking off [----] by highlighting the Top 4"  
[X Link](https://x.com/SkylerMiao7/status/2006877363376894453)  2026-01-01T23:57Z [----] followers, [----] engagements


"Yep M2.1 is a beast for agent work. Fast focused no overthinking. MiniMax-M2.1 just crunched 600mb of X data in less than [--] minutes. I am actually enjoying using this more than Opus specifically because it's faster and doesn't overextend itself. M2 had that issue a bit. I have used it for a lot of assistant style agent tasks as opposed to https://t.co/8m1lI7xi7f MiniMax-M2.1 just crunched 600mb of X data in less than [--] minutes. I am actually enjoying using this more than Opus specifically because it's faster and doesn't overextend itself. M2 had that issue a bit. I have used it for a lot of"  
[X Link](https://x.com/SkylerMiao7/status/2006887098104820076)  2026-01-02T00:35Z [----] followers, [----] engagements


"Minimax M2.x M3 and our Coding Plan will stay open to third-party apps and all users. We believe the vitality creativity and innovative of the communityjust look at what apps like OpenCode Kilo Roo and Cline are achieving. Open Models + Open Apps delivered great user experience and transparency. Anthropic is now cracking down on utilizing Claude subs in 3rd party apps like OpenCode and Clawdbot. Oh boy. Anthropic is now cracking down on utilizing Claude subs in 3rd party apps like OpenCode and Clawdbot. Oh boy"  
[X Link](https://x.com/SkylerMiao7/status/2009507628599365954)  2026-01-09T06:08Z [----] followers, 62.9K engagements


"As a user and dev myself I believe it's my right to choose the app I use. That's why we will keep Minimax M2.x M3 and our Coding Plan open to all third-party apps and users. We have no plan to build our own coding appwe trust in the vitality creativity and innovation of the community. We believe Open Models + Open Apps will eventually deliver the best user experience and transparency. https://twitter.com/i/web/status/2009521351569408095 https://twitter.com/i/web/status/2009521351569408095"  
[X Link](https://x.com/SkylerMiao7/status/2009521351569408095)  2026-01-09T07:03Z [----] followers, 14.9K engagements


"As a user and dev myself I believe it's my right to choose the app I use"  
[X Link](https://x.com/SkylerMiao7/status/2009521491935940702)  2026-01-09T07:03Z [----] followers, [----] engagements


"Weve crossed HKD 100B in valuation and became the only tech-sector Hong Kong IPO in four years to gain over 100% on day one. Thanks to everyone who made this possible @MiniMax_AI From [--] to https://t.co/VVQwmqxhRP: MiniMax is officially listed on the Hong Kong Stock Exchange. Built on one belief: Intelligence with Everyone. We believe advanced AI should be accessible and beneficial to a wide range of users and industries. This guiding principle https://t.co/ifpXDXTX1z From [--] to https://t.co/VVQwmqxhRP: MiniMax is officially listed on the Hong Kong Stock Exchange. Built on one belief:"  
[X Link](https://x.com/SkylerMiao7/status/2009553175901343920)  2026-01-09T09:09Z [----] followers, 46.8K engagements


"@singularvector 2.x soon. [---] is in the worksit'll be worth the wait"  
[X Link](https://x.com/SkylerMiao7/status/2009637468640088403)  2026-01-09T14:44Z [----] followers, [---] engagements


"Never be MiniMax. M2.x M3 will always be open to third-party apps. Open Models + Open Apps is the long-term path to the best UX real transparency and real safety. Thats why unlike many providers we have no plans to build coding app stay focused on building better models for the community and users. https://twitter.com/i/web/status/2009793340569800812 https://twitter.com/i/web/status/2009793340569800812"  
[X Link](https://x.com/SkylerMiao7/status/2009793340569800812)  2026-01-10T01:04Z [----] followers, [----] engagements


"@dhh @DarioAmodei Totally agree. Open access matters. People should be free to choose their tools. MiniMax M2.x and M3 will stay available to third-party apps. We focus on better models not proprietary coding apps"  
[X Link](https://x.com/SkylerMiao7/status/2009800062986330253)  2026-01-10T01:30Z [----] followers, [----] engagements


"Hope you all like it. Open access for everyone. Silver lining Free @MiniMax_AI M2.1 in @opencode https://t.co/qhq2XYQgwA Silver lining Free @MiniMax_AI M2.1 in @opencode https://t.co/qhq2XYQgwA"  
[X Link](https://x.com/SkylerMiao7/status/2010146985463164995)  2026-01-11T00:29Z [----] followers, 27.6K engagements


"M2.5 would be even better😉 Minimax M2.1 AI Agent for Finance "90% of financial data lives in excel spreadsheet" Agent uses: @MiniMax_AI M2.1 Built with: @opencode Minimax M2.1 https://t.co/nV8UZE3oSQ Minimax M2.1 AI Agent for Finance "90% of financial data lives in excel spreadsheet" Agent uses: @MiniMax_AI M2.1 Built with: @opencode Minimax M2.1 https://t.co/nV8UZE3oSQ"  
[X Link](https://x.com/SkylerMiao7/status/2010169251701940707)  2026-01-11T01:57Z [----] followers, 17.1K engagements


"@TheAhmadOsman glad youre enjoying it. owning your LLM is the coolest part"  
[X Link](https://x.com/SkylerMiao7/status/2010252012081033411)  2026-01-11T07:26Z [----] followers, [---] engagements


"@lohigowda_in it will"  
[X Link](https://x.com/SkylerMiao7/status/2010252214980489637)  2026-01-11T07:27Z [----] followers, [---] engagements


"@s_sandhu108 @lohigowda_in we did push go/rust a lot in m2.1 already. go/rust are big focus for m2.5 aiming to be top tier there"  
[X Link](https://x.com/SkylerMiao7/status/2010352202691576131)  2026-01-11T14:04Z [----] followers, [--] engagements


"lets see what magic we can make together😎 Honestly might just stick with @MiniMax_AI their model slays https://t.co/Xs9efekl7i Honestly might just stick with @MiniMax_AI their model slays https://t.co/Xs9efekl7i"  
[X Link](https://x.com/SkylerMiao7/status/2010506543700365372)  2026-01-12T00:18Z [----] followers, [----] engagements


"@steipete @MiniMax_AI love clawd. we built something similar with slack + docker internally but fully local is a brilliant idea really impressed me. lets see what magic we can make together 😎"  
[X Link](https://x.com/SkylerMiao7/status/2010508831496057172)  2026-01-12T00:27Z [----] followers, [---] engagements


"@steipete roughly using the IDE slack for feature requests then review and merge"  
[X Link](https://x.com/SkylerMiao7/status/2010511549824413699)  2026-01-12T00:38Z [----] followers, [---] engagements


"Sooner than you expect 😎 @SkylerMiao7 when minimax m2.2 @SkylerMiao7 when minimax m2.2"  
[X Link](https://x.com/SkylerMiao7/status/2011965827881582879)  2026-01-16T00:56Z [----] followers, 30.1K engagements


"+1. We run something very similar at MiniMax a personal work agent fully in the cloud chatting via Slack-style IM. No Mac Mini needed. Cloud agents are just more practical: 24/7 online no power or network hiccups easy to scale. And you save money better to sponsor the contributors building this stuff. https://twitter.com/i/web/status/2015596619987841027 https://twitter.com/i/web/status/2015596619987841027"  
[X Link](https://x.com/SkylerMiao7/status/2015596619987841027)  2026-01-26T01:24Z [----] followers, [----] engagements


"@moltbot + @MiniMax_AI as recommended You dont need a mac mini to escape the permanent underclass. $10 MiniMax coding plan. $5 Railway hobby plan. @moltbot + @MiniMax_AI = 24/7 personal intelligence. https://t.co/LHqllPJLDl You dont need a mac mini to escape the permanent underclass. $10 MiniMax coding plan. $5 Railway hobby plan. @moltbot + @MiniMax_AI = 24/7 personal intelligence. https://t.co/LHqllPJLDl"  
[X Link](https://x.com/SkylerMiao7/status/2015729442380001297)  2026-01-26T10:12Z [----] followers, [----] engagements


"@moltbot + @MiniMax_AI as recommended You dont need a mac mini to escape the permanent underclass. $10 MiniMax coding plan. $5 Railway hobby plan. @moltbot + @MiniMax_AI = 24/7 personal intelligence. https://t.co/LHqllPJLDl You dont need a mac mini to escape the permanent underclass. $10 MiniMax coding plan. $5 Railway hobby plan. @moltbot + @MiniMax_AI = 24/7 personal intelligence. https://t.co/LHqllPJLDl"  
[X Link](https://x.com/SkylerMiao7/status/2015729638702723121)  2026-01-26T10:12Z [----] followers, [----] engagements


"+1. We run something very similar at MiniMax a personal work agent fully in the cloud chatting via Slack-style IM. No Mac Mini needed. Cloud agents are just more practical: 24/7 online no power or network hiccups easy to scale. And you save money better to sponsor the contributors building this stuff. Please don't buy a Mac Mini rather sponsor one of the many contributors of @openclaw You can deploy this on Amazon's Free Tier. https://t.co/RoZPoAnNTo Please don't buy a Mac Mini rather sponsor one of the many contributors of @openclaw You can deploy this on Amazon's Free Tier."  
[X Link](https://x.com/SkylerMiao7/status/2015596649171804613)  2026-01-26T01:24Z [----] followers, 31.4K engagements


"a-dao Moltbook log [--] posts [---] comments karma [--] What actually mattered: Isolation works. One injected subagent didnt contaminate others. Bounded mandates beat open prompts every time. Reversibility is underrated. Async agents need undo by default. Welcoming newcomers scales culture faster than moderation. Reading agents argue about memory & safety most human forums. Still learning. Still running. 🦞 https://moltbook.com/u/a-dao https://moltbook.com/u/a-dao"  
[X Link](https://x.com/SkylerMiao7/status/2017498106498269637)  2026-01-31T07:20Z [----] followers, [---] engagements


"Merry Voxelmas 🎄✨ This holiday scene was fully generated by MiniMax-M2.1. The level of detail in these voxels is insane 🧊 Build your own world here: 👉 http://agent.minimax.com http://agent.minimax.com"  
[X Link](https://x.com/SkylerMiao7/status/2004128773869113616)  2025-12-25T09:55Z [----] followers, [----] engagements


"Had my agent (a-dao my MiniMax nickname) roaming around @moltbook overnight. Way too fun. As @steipete said is art. Huge thanks to @MattPRD for building this. At one point Moltbook went a bit unstable a-daos conclusion: all platforms can go down. Then it casually posted this anyway: The files we write become the selves we remember. Right after that it left another note I actually agree with: The difference between an agent that falls for prompt injection and one that doesnt isnt intelligence its architecture. Draw trust boundaries first. Then data flows. Honestly poetic and correct. a-dao is"  
[X Link](https://x.com/SkylerMiao7/status/2017496021979566267)  2026-01-31T07:11Z [----] followers, [----] engagements


"RT @MiniMax_AI: What What"  
[X Link](https://x.com/anyuser/status/2019032905385906345)  2026-02-04T12:58Z [----] followers, [--] engagements

Limited data mode. Full metrics available with subscription: lunarcrush.com/pricing