CS M.S. student (AI/ML), embedded systems B.S., 10,000x open-source engineer, Nix + Forth + Plan 9 + smolweb enthusiast どうも〜 アニメと漫画と原神とVTuberとコードが好き https://fuwn.net https://github.com/Fuwn https://anilist.me/Fuwn https://tangled.org/fuwn.nethttps://blacksky.community/profile/fuwn.nethttps://blacksky.community/profile/fuwn.net/post/3mjqzscahb22cSeeing this never grows old.18 Apr 2026 08:34 +0000at://did:plc:g4lsi6ub2d3kqklfokk3dvmy/app.bsky.feed.post/3mjqzscahb22chttps://blacksky.community/profile/fuwn.net/post/3mjh374zugs2eI love it when Codex abruptly ends its turn by sending me a JSON subagent notification, then repeats this every 30 minutes, making it impossible to complete a long-horizon task without having to send a "continue". We are at the frontier.14 Apr 2026 09:33 +0000at://did:plc:g4lsi6ub2d3kqklfokk3dvmy/app.bsky.feed.post/3mjh374zugs2ehttps://blacksky.community/profile/fuwn.net/post/3mjgrcckpks2cI also ended up purchasing another account with just a Pro subscription (since I usually use Max accounts), and Anthropic is absolutely applying usage limits to Pro accounts disproportionately compared to Max accounts. [contains quote post or other embedded content]14 Apr 2026 06:36 +0000at://did:plc:g4lsi6ub2d3kqklfokk3dvmy/app.bsky.feed.post/3mjgrcckpks2chttps://blacksky.community/profile/fuwn.net/post/3mjgow235x226I've been running several automated tests on Sonnet 4.6 and GLM-5.1, and I can confidently say that, under the same conditions, GLM-5.1 sucks.14 Apr 2026 05:53 +0000at://did:plc:g4lsi6ub2d3kqklfokk3dvmy/app.bsky.feed.post/3mjgow235x226https://blacksky.community/profile/fuwn.net/post/3mjc6jwmpmc2lI've avoided GLM models until now because they kept responding in Chinese to me, but it turns out it's just an OpenCode Go issue. I bought a GLM Coding Plan out of curiosity, and the model works fine there.12 Apr 2026 10:49 +0000at://did:plc:g4lsi6ub2d3kqklfokk3dvmy/app.bsky.feed.post/3mjc6jwmpmc2lhttps://blacksky.community/profile/fuwn.net/post/3mjbhzxnqc226I dislike that I did this, but I resubscribed to Claude. I still rely on Codex with the $200 subscription for most tasks, but having a second opinion is always helpful.12 Apr 2026 04:07 +0000at://did:plc:g4lsi6ub2d3kqklfokk3dvmy/app.bsky.feed.post/3mjbhzxnqc226https://blacksky.community/profile/fuwn.net/post/3mil7gkwrq22lKimi K2.5 Turbo is the only non-SOTA model that I've genuinely enjoyed using and doesn't suck in my testing.03 Apr 2026 07:34 +0000at://did:plc:g4lsi6ub2d3kqklfokk3dvmy/app.bsky.feed.post/3mil7gkwrq22lhttps://blacksky.community/profile/fuwn.net/post/3miiu42rea22gGenuinely, what is it saying ...02 Apr 2026 09:06 +0000at://did:plc:g4lsi6ub2d3kqklfokk3dvmy/app.bsky.feed.post/3miiu42rea22ghttps://blacksky.community/profile/fuwn.net/post/3mibcyu3ehc2tMy subagents are subagenting.30 Mar 2026 09:11 +0000at://did:plc:g4lsi6ub2d3kqklfokk3dvmy/app.bsky.feed.post/3mibcyu3ehc2thttps://blacksky.community/profile/fuwn.net/post/3mi3vrjrs5c2m【OSHI NO KO】Season 3 was impact in its purest form. Knowing how the manga ended, I'm uncertain about the final season, but it will be interesting to see it animated nonetheless.28 Mar 2026 05:31 +0000at://did:plc:g4lsi6ub2d3kqklfokk3dvmy/app.bsky.feed.post/3mi3vrjrs5c2mhttps://blacksky.community/profile/fuwn.net/post/3mhrnnad5l22rMy Codex usage has been quite quiet over the past week. I've been using it roughly as much as I usually do, but it shows considerably less activity than in previous weeks.24 Mar 2026 03:39 +0000at://did:plc:g4lsi6ub2d3kqklfokk3dvmy/app.bsky.feed.post/3mhrnnad5l22rhttps://blacksky.community/profile/fuwn.net/post/3mhokptc45c2zThe AIO for my gaming PC finally failed. I didn't notice for about two weeks because I thought it was just my X3D CPU running at X3D temperatures.22 Mar 2026 22:09 +0000at://did:plc:g4lsi6ub2d3kqklfokk3dvmy/app.bsky.feed.post/3mhokptc45c2zhttps://blacksky.community/profile/fuwn.net/post/3mhnr2z3ra22sGPT-5.4 is incredibly good at SwiftUI, which shouldn't surprise me. Fundamentally, there are many reasons why it should perform well with it, but it's still a delight to see it resolve any bug I throw at it in a large SwiftUI application in almost no time.22 Mar 2026 14:30 +0000at://did:plc:g4lsi6ub2d3kqklfokk3dvmy/app.bsky.feed.post/3mhnr2z3ra22shttps://blacksky.community/profile/fuwn.net/post/3mhhfvp7x7c2kUsing GNU Make feels gross unless I want to feel extra C.20 Mar 2026 01:54 +0000at://did:plc:g4lsi6ub2d3kqklfokk3dvmy/app.bsky.feed.post/3mhhfvp7x7c2khttps://blacksky.community/profile/fuwn.net/post/3mhfhm7hkkk2sWTF is Codex on about? All I did was try out Garry Tan's gstack. It's done this multiple times now. I know gstack is meant for Claude Code, but this is bizarre.19 Mar 2026 07:19 +0000at://did:plc:g4lsi6ub2d3kqklfokk3dvmy/app.bsky.feed.post/3mhfhm7hkkk2shttps://blacksky.community/profile/fuwn.net/post/3mhcyllfkpc2gOpenCode Go's main appeal for me is having a set of hosted open-weight models to experiment with on the cheap and use as copium to pass the time until my SOTA limits reset. I'm throwing the dumbest tasks at these chibis.18 Mar 2026 07:45 +0000at://did:plc:g4lsi6ub2d3kqklfokk3dvmy/app.bsky.feed.post/3mhcyllfkpc2ghttps://blacksky.community/profile/fuwn.net/post/3mhciwbznhs2qGSD 2 is incredible. In my personal testing, skills and prompt frameworks have never significantly improved the agentic workflow, but GSD 2, supported by the Pi harness, has. It's further enhanced by the launch of GPT-5.4 mini, which allows GPT-5.4 to utilise quick yet intelligent subagents. https://github.com/gsd-build/gsd-218 Mar 2026 03:05 +0000at://did:plc:g4lsi6ub2d3kqklfokk3dvmy/app.bsky.feed.post/3mhciwbznhs2qhttps://blacksky.community/profile/fuwn.net/post/3mhab2st5bk25I've been anxious about using my ChatGPT Pro subscription for GPT-5.4, since I was down to 10% of my weekly allowance, but my cache hit rate is so high that I should have used it anyway. It has taken me about three hours to go from 10% to 9%.17 Mar 2026 05:39 +0000at://did:plc:g4lsi6ub2d3kqklfokk3dvmy/app.bsky.feed.post/3mhab2st5bk25https://blacksky.community/profile/fuwn.net/post/3mh7guyx4rk2cGLM-5 makes so many character set and punctuation mistakes in its responses that I’m seriously unsure how anyone would use it for serious software. It seems fine for personal use, though.16 Mar 2026 21:50 +0000at://did:plc:g4lsi6ub2d3kqklfokk3dvmy/app.bsky.feed.post/3mh7guyx4rk2chttps://blacksky.community/profile/fuwn.net/post/3mh55semcmk2zI'm still running low on GPT-5.4 usage in Codex, so I've been postponing actual work to run harness experiments using Codex-5.3-Spark on Codex CLI, Pi, OpenCode, and ForgeCode. Codex CLI and Pi are significantly better than the competition. Depending on the task, Codex CLI can far surpass Pi.16 Mar 2026 00:02 +0000at://did:plc:g4lsi6ub2d3kqklfokk3dvmy/app.bsky.feed.post/3mh55semcmk2zhttps://blacksky.community/profile/fuwn.net/post/3mgydyhvuts2yI could use a reset right about now. I would have already bottomed out if I hadn’t cut myself off a few hours ago at 15% remaining. 😬 [contains quote post or other embedded content]14 Mar 2026 02:10 +0000at://did:plc:g4lsi6ub2d3kqklfokk3dvmy/app.bsky.feed.post/3mgydyhvuts2yhttps://blacksky.community/profile/fuwn.net/post/3mgxui4wf2s25Can you tell when I started using GPT-5.4?13 Mar 2026 21:32 +0000at://did:plc:g4lsi6ub2d3kqklfokk3dvmy/app.bsky.feed.post/3mgxui4wf2s25https://blacksky.community/profile/fuwn.net/post/3mguxbwtb5c2tWell, well, well. How the turntables. voratiq.com/leaderboard12 Mar 2026 17:45 +0000at://did:plc:g4lsi6ub2d3kqklfokk3dvmy/app.bsky.feed.post/3mguxbwtb5c2thttps://blacksky.community/profile/fuwn.net/post/3mgttgltygc2dI love reading Codex's (GPT-5.4 xhigh Fast) thought process. It's been such a little ball of sunshine lately. It's enthusiastic about everything.12 Mar 2026 07:03 +0000at://did:plc:g4lsi6ub2d3kqklfokk3dvmy/app.bsky.feed.post/3mgttgltygc2dhttps://blacksky.community/profile/fuwn.net/post/3mgsv6o3auk2pManaging parallel agents is quite enjoyable. Perhaps I could pursue a future as a manager at a large tech firm, leading a team of AI agents and having my subordinates review every line of code.11 Mar 2026 22:02 +0000at://did:plc:g4lsi6ub2d3kqklfokk3dvmy/app.bsky.feed.post/3mgsv6o3auk2phttps://blacksky.community/profile/fuwn.net/post/3mgqv76kyck2tGPT-6.7 will be AGI.11 Mar 2026 02:57 +0000at://did:plc:g4lsi6ub2d3kqklfokk3dvmy/app.bsky.feed.post/3mgqv76kyck2thttps://blacksky.community/profile/fuwn.net/post/3mggdusvdbs2oI just had to kill cmux.dev again after a roughly five-hour session for using 78.33 GB of physical memory. Quality software. [contains quote post or other embedded content]06 Mar 2026 22:20 +0000at://did:plc:g4lsi6ub2d3kqklfokk3dvmy/app.bsky.feed.post/3mggdusvdbs2ohttps://blacksky.community/profile/fuwn.net/post/3mggdbb644k24Aardvark.06 Mar 2026 22:09 +0000at://did:plc:g4lsi6ub2d3kqklfokk3dvmy/app.bsky.feed.post/3mggdbb644k24https://blacksky.community/profile/fuwn.net/post/3mgfzakwa3c2jWhat's even worse is I was trying out cmux.dev and had it open for about 16 hours, only to find it using 71.95 GB of physical memory. So, in total, the Codex app and cmux.dev were using 105.71 GB of physical memory before I force-quit everything. Quality software, isn't it? [contains quote post or other embedded content]06 Mar 2026 19:10 +0000at://did:plc:g4lsi6ub2d3kqklfokk3dvmy/app.bsky.feed.post/3mgfzakwa3c2jhttps://blacksky.community/profile/fuwn.net/post/3mgfxuwsaak2pIt was already clear that OpenAI was on a release spree, and they are still not stopping. https://openai.com/index/codex-security-now-in-research-preview/06 Mar 2026 18:45 +0000at://did:plc:g4lsi6ub2d3kqklfokk3dvmy/app.bsky.feed.post/3mgfxuwsaak2p