I was very happy that you guys liked the idea!
This is a dev post classified by AGK Intelligence as AI dev tools (a launch), kept by the Dev Radar because it carries real work, not commentary.
I was very happy that you guys liked the idea! Here is the first real battle: Qwen3.8-27B on my RTX 5090 vs. DeepSeek-V4.1-Flash on my Mac Studio. RTX 5090: Qwen3.8-27B NVFP4 and FP8 mix, NInfer engine, MTP 3 Mac Studio M5 Ultra: DeepSeek-V4.1-Flash-1017 Bob's TensorFold Metal recipe MLX, experts at 2 and 3 bits This is the strategy game I'm building. It runs in real time. Both models think before they act, no human input. DeepSeek runs greedy as in Bob's recipe. Qwen runs at temperature 1.0 as on its model card. DeepSeek scores higher on my knowledge and code tests. Qwen decodes faste
Posted by Niklas (2.7k followers) 1 days ago · 149 likes · 10.4k views · view the original post on X. Kept by the Dev Radar as AI dev tools. Tools mentioned: niklaslenz.com.
More dev work like this
- (1/6) @pika_labs open-sourced @MiniMax_AI H3 Flex late last week. Today it's in… — @haoailab
- GLM new model? — @0x0SojalSec
- Turn one script into a narrated video — without assembling every asset by hand. — @DanKornas
- Oh this looks interesting. — @ItsmeAjayKV
- OCRia-1 is one of the fastest models ever for doc parsing. We sped up OCRia-1 by 10x on… — @jerryjliu0
- Pra que não sabia, eu tenho meu próprio launcher de jogos: Omarchy Games — @AkitaOnRails
- 65 new 𝗛𝗲𝗿𝗺𝗲𝘀 plugins joined the catalog since Monday. These 5 fix problems you… — @BkashJosi
- Tavily is now available on @Composio Instant. — @tavilyai
Every post is read and classified by AGK Intelligence: what it is, which market it belongs to, and whether the link is a real tool. 31.2k posts from 5k X accounts over the last 21 days, 9.2k tools, 12 markets. New posts are collected several times an hour and the full ranking is recomputed every hour — last update 2026-11-03 20:22 UTC. Full method.