- Agent coding addict : https://teapaw.org/t.co/ZwbyJ8ZbKQ - Co-organizer, Vibe Coding Community Paris : https://teapaw.org/t.co/y20yrcs1nn
Paris
Joined September 2009
- Tweets3.2K
- Following926
- Followers352
- Likes910
Pinned Tweet
🐸 I just released opencode-froggy, a plugin for @opencode bringing #claudecode style hooks on top of #opencode.
```
hooks:
- event: session.idle
conditions: [hasCodeChange]
actions:
- bash: "npm run lint --fix"
```
github.com/smartfrog/opencod…
#videcoding #aicode
Celeste on PS Vita 🏔️
Started working on this port last night! Here’s a first look at it running on real hardware.
Still a work in progress, with a lot of work ahead!
#Celeste #PSVita #VitaHomebrew
1
2
349
Paris Bench 🇫🇷 - DeepSeek 4.1 Flash Max
Ista IQ2_XS · via Project Maya · 33 tok/s ⚡
Nice speed, but the quality loss from quantization is noticeable.
Excited for Qwen4! 🤞 Smaller models would let us run higher-quality quants ✨
cc @PeasantSmith
Qwen 4 Preview — Paris at Night Bench 🌙
Same photo, same prompt: recreate Paris in 3D with Three.js and a cinematic drone camera.
Qwen 4 Preview (Max) vs DeepSeek Flash 4.1(High).
Which one do you prefer?
2
2
679
Qwen 4 Preview — Paris at Night Bench 🌙
Same photo, same prompt: recreate Paris in 3D with Three.js and a cinematic drone camera.
Qwen 4 Preview (Max) vs DeepSeek Flash 4.1(High).
Which one do you prefer?
Qwen 4 Preview : PICO‑8/Celeste Bench 🎮🏔️
📝 Same initial prompt:
🟣 Qwen 4 Preview xhigh: 46m40s · 12.59M tokens
🔵 GLM‑5.3 high: 53m34s · 12.89M tokens
🟢 GPT‑6.1 Sol xhigh: 12m22s · 3.65M tokens
117
Maya 1.1.54 vs Strata 0.2.11 🥊
2×RTX 3090 · 128GB RAM
DeepSeek V4.1 Flash · Unsloth UD-IQ1_M
Median decode:
Maya: 26.04 tok/s
Strata: 22.75 tok/s
Strata vs Maya:
+19.3% read time
+21.8% total time
Any config tips, @coldniko @PeasantSmith?
The shared setup 🔧
Split 14/26 · TP1 · chunk4096 · context200k
CPU8 · reserve512 MiB
Engram: 256 MiB cache, 32 readers
Full pinned expert RAM copies
6-expert routing · no DSpark
Upstream native agent bench:
900s each · 180s warmup · 1 client
1
144
Fred D. | 一铭 retweeted
Today, we release the weights of a new and improved Le Chonk, advancing the open frontier for agentic coding and cybersecurity.
💻 Built to build: Le Chonk ranks #1 on @cognition’s FrontierCode among open models, and sets a new benchmark for Pareto efficiency on @artificialanalysis.
🛡 Built to defend: State-of-the-art in cyberdefense, put through the paces by the world's leading cyber vendors, including @CrowdStrike, @cloudflare, @rubrik, @zscaler, and others.
Built for sovereignty: Make it yours. Download the weights on @huggingface , customize it with Mistral Forge, and deploy it wherever you need it.
Read more from our team and customers 👇:
Fred D. | 一铭 retweeted
🚀 Introducing DeepInfer, our in-house inference framework, now fully open source!
Powering DeepSeek-V4.1-Flash at scale with high throughput and low latency. We’re sharing our production stack to help the community lower compute costs, reduce energy use, and make AI more affordable.
🔗 github.com/deepseek-ai/DeepI…
1/4
Fred D. | 一铭 retweeted
Sylvia Serfati sur la révolution de l'IA en mathématiques
"La plupart des résultats d'OpenAI mériteraient la médaille Fields".
"Les preuves sont d'une complexité incroyable et aussi parfois d'une beauté impressionnante. C'est-à-dire que nous n'avons pas le monopole de la beauté."
Passionnante réflexion des mathématiciens @Amaury_Hayat et Sylvia Serfati sur l'avenir de leur profession à l'ère de l'IA : la résolution de problèmes ne sera plus notre rôle, et poser les bonnes questions, sans doute que même l'IA le fera mieux que nous à terme...
Fred D. | 一铭 retweeted
Kaiko's Data On-Ramp oracle is the first institutional-grade oracle on @ZenithFdn , live on testnet now. Ethereum Virtual Machine (EVM) developers building on @CantonNetwork can access regulated, multi-asset pricing data ahead of mainnet launch.
Link to the full announcement 🔗 kaiko.com/news/kaiko-deploys…
