@ashen_one @XFreeze Lmao it’s so ass, can’t believe he posted it lmao
English
Jamthelion 🐝
3.3K posts
checkpoint of image v2 is rolling out to a lot of users
i hope we get a better version later
Leon Lin@LexnLin
my last guess: Today superapp Tuesday image v2 next Thursday GPT 5.5
English
English
@LexnLin @thegenioo Yeah but the dense 27B models are pretty cracked for its size. Considering you can run it on a single gpu.
English
benchmaxxed + they compare it to opus 4.5 (not .6/.7 lol)
Qwen@Alibaba_Qwen
🚀 Introducing Qwen3.6-Max-Preview, an early preview of our next flagship model Highlights： ⚡️ Improved agentic coding capability over Qwen3.6-Plus 📖 Stronger world knowledge and instruction following 🌍 Improved real-world agent and knowledge reliability performance Smarter, sharper, still evolving. More Qwen3.6 models to come. Stay tuned! 🔗👇 Blog: qwen.ai/blog?id=qwen3.… Qwen Studio: chat.qwen.ai/?models=qwen3.… API: modelstudio.console.alibabacloud.com/ap-southeast-1…
English
wtf, did they take the 5.5 Pro back?
Now I got this one for the same prompt
Leon Lin@LexnLin
gave 5.5 pro a reference image and it did this :)
English
@Angaisb_ 100% but icl some have been spot on. I also do think GPT 5.4 Pro got an upgrade of some sort possibly GPT 5.5 testing or what not. Just take a look at this. I was skeptical at first but after testing it 100% got upgraded.
x.com/LexnLin/status…
Leon Lin@LexnLin
gave 5.5 pro a reference image and it did this :)
English
can't help but notice every new chinese model's benchmark visuals seem to compare to previous generations
opus 4.7 conspicuously absent.. lol
Kimi.ai@Kimi_Moonshot
Meet Kimi K2.6: Advancing Open-Source Coding 🔹Open-source SOTA on HLE w/ tools (54.0), SWE-Bench Pro (58.6), SWE-bench Multilingual (76.7), BrowseComp (83.2), Toolathlon (50.0), Charxiv w/ python(86.7), Math Vision w/ python (93.2) What's new: 🔹Long-horizon coding - 4,000+ tool calls, over 12 hours of continuous execution, with generalization across languages (Rust, Go, Python) and tasks (frontend, devops, perf optimization). 🔹Motion-rich frontend - Videos in hero sections, WebGL shaders, GSAP + Framer Motion, Three.js 3D. 🔹Agent Swarms, elevated - 300 parallel sub-agents × 4,000 steps per run (up from K2.5's 100 / 1,500). One prompt, 100+ files. 🔹Proactive Agents - K2.6 model powers OpenClaw, Hermes Agent, etc for 24/7 autonomous ops. 🔹Claw Groups (research preview) - bring your own agents, command your friends', bots & humans in the loop. - K2.6 is now live on kimi.com in chat mode and agent mode. For production-grade coding, pair K2.6 with Kimi Code: kimi.com/code - 🔗 API: platform.moonshot.ai 🔗 Tech blog: kimi.com/blog/kimi-k2-6 🔗 Weights & code: huggingface.co/moonshotai/Kim…
English
@Jamthelion That isn't how it works anymore. It's about creating a model that is incrementally improved. Then, when it's ready, you feed it to Claude Design to figure out how to design the UI. Totally different method.
English
Mind-blown 🤯. Claude Design allows one to build complex interfaces that front complex ontologies in a flash! Having a well-defined domain model that drives the design yields an unfair speed advantage. This accelerates any QPT-derived model! Palantir is going to be challenged. @claudeai @ClaudeDevs @AnthropicAI
English
just used your reply as the prompt :)
Michael Goldstein@mlg27_
@LexnLin Make a fully working 3d flight sim using threeJS
English
@LexnLin Its crazy, just tested the voxel pelican, tbh tho tried building a front end design on it and icl codex with GPT 5.4 Xhigh smoked it. In less time too. So im not sure if the prompt just didnt trigger the 5.5 model.
English
@techikansh This, my old macbook M1 with 8 gb ram is struggling so bad lol.
English
8 GB of RAM is NOT enough.
It may outperform Windows in some cases, but don’t buy the base MacBook Neo if you’re serious about coding.
Andres Vidoza@andres_vidoza
The Cheapest MacBook… Is It Enough for Coding?
English
@LexnLin Is it GPT 5.4 Pro with standard thinking or are you using extended?
English
🚨 Here we go again
GPT PRO UPDATE amazes me like crazy
gave image and tell it to code the game with same aesthetic it coded it in 10 min
iMuffin@iMuffined
@chetaslua @OpenAI @sama "make a instantly playable detailed 3d office punch-em-up game"
English
@synthwavedd @anko_979 Something def changed, its much quicker, same exact prompt took me 39 mins 2 days ago. It just took me 11 minutes with arguably better output.
English
spud is 5.5, and it will be significantly better than 5.4, but think of it as a half cooked version of gpt-6 (which will be spud at its full potential!)
Haider.@haider1
still unclear why people call "spud" gpt-5.5 when it is a completely new pre-trained model if spud is much better than gpt-5.4 and not just a small upgrade, then calling it gpt-5.5 seems very odd it makes more sense to call it gpt-6 gpt-5.5 and spud should be different models
English
