Stanford just dropped a full course on self-improving AI agents.
The professors built Claude and Gemini. "we sampled the model 10,000 times. only 3 answers were correct. but those 3 solved what the best models in the world couldn't." 0:00 – scaling laws: why bigger models keep
Deepseek-v4-flash-abliterated seems to be such a chill guy afterall.
Is this what Anthropic is fearmongering 🤣 Told it to generate one short video using minimax h3 and a short music sample using minimax music3 and use its own taste, then connect them together. To me, it
I'm using deepseek through opencode too, really free plan.
In two weekends, I ported this app that was in electron to tauri 2. implemented some new features, removed a lot of unnecessary stuff, etc. At zero cost, crazy.
Fable 5 vs DeepSeek V4 Pro vs Grok 4.6: the number nobody's talking about is 276.
Not the intelligence score. The cache read gap. Agents re-read your instructions on EVERY step. Hundreds of times per task. DeepSeek's cache reads cost ~276x less than Fable 5's. With a 92% hit
You have reached the end of the archive
All of deepseek