Tag: Open source
-
+
+
+
+
+
+
Chinese startup MiniMax just dropped M2.5, its most powerful model yet. But M2.5 flopped in my testing, scoring a C+. That’s only slightly better than MiniMax’s prior model, which I gave a C last month. Let me show you where this model works and where it fails… Round #1: Automating…
+
+
+
+
+
+
+ AI, Artificial Intelligence, ChatGPT, China, LLM, MiniMax, Open source, Technology, Venture Capital, Writing+
+
+
+
+
+
+
+
+
+
+
+
-
+
+
+
+
+
+
Moonshot AI recently dropped Kimi K2.5, its most powerful model ever. But in my testing, K2.5 failed miserably. When I reviewed Kimi K2 in November, it was a real threat to ChatGPT and Grok. But K2.5 feels like a major downgrade. Across a range of queries, K2.5 delivered useless results.…
+
+
+
+
+
+
+ AI, Artificial Intelligence, ChatGPT, China, DeepSeek, Investing, Kimi, LLM, Open source, Startups, Technology, Venture Capital+
+
+
+
+
+
+
+
+
+
+
+
-
+
+
+
+
+
+
The Allen Institute for AI recently released its most powerful model ever: Olmo 3. Olmo is cheap to train, making it perfect for anyone training their own model. Olmo 3 is 2.5 times more efficient to train than Llama 3.1 based on GPU-hours per token. Olmo is also much more…
+
+
+
+
+
+
+ AI, Artificial Intelligence, ChatGPT, LLM, Model Training, Open source, OSS, Tech, Technology, Venture Capital+
+
+
+
+
+
+
+
+
+
+
+
-
Kimi just dropped Kimi 2 Thinking. I tested it this morning. It’s as good as Grok 4, the best model I’ve ever used. In July, I ran the prior Kimi through testing. It scored a B+, solid but below Grok 4. This time, Kimi performed far better. Let me show…
+
+
+
+
+
+
+ AI, Artificial Intelligence, ChatGPT, China, Kimi, LLM, Open source, Startups, Technology, Venture Capital, Writing+
+
+
+
+
+
+
+
+
+
+
+
Tremendous 