Mistral announced a new mistral-small-creative 32k model on their Discord! An experimental small model designed for creative writing, roleplay and character-driven dialogue, general-purpose instruction following, and conversational agents.
LLMS
-
Foundation Model Transparency Index 2025 Shows Sharp Decline
By
–
The 2025 Foundation Model Transparency Index is here! This report reveals a sharp drop in openness, with average scores plummeting from 58 to 40. While IBM leads with 95, xAI and Midjourney scored just 14. – Google: 41
– OpenAI: 35
– DeepSeek: 32
– Qwen: 26 The 2025 -
Error-Free Linear Attention Surpasses Hybrid Attention Methods
By
–
With the recent hybrid attention releases from MiniMax, Qwen, Kimi, and NVIDIA, this paper introduces Error-Free Linear Attention that could top them all This new technique has a stable linear-time attention that's better than any linear attention variants and also DeltaNet!
-
Does AI Have Imagination? Technical Facts Explained
By
–
Does AI have imagination?
This seems to be the debate my original post has sparked. My take is: yes, but not in the way human beings imagine. I’m presenting the technical facts below. You can arrive at your own conclusion. When people say “AI has no imagination,” they’re -
AI imagination debate: technical nuances and probabilistic generation
By
–
Since this is sparking a debate, and many people are claiming that AI has no imagination – here are some technical nuances. When people say “AI has no imagination,” they’re usually confusing experience with capability. Technically, AI generates stories using probabilistic
-
GPT Image 1.5 tops LM Arena
By
–
GPT Image 1.5 from OpenAI took the first spot on LM Arena in text to image category with a 29 points gap. Competition
-
GPT-1.5 vs Nano-Banana Pro: Performance and Prompt Engineering Comparison
By
–
My anecdotal impression of GPT-1.5 vs Nano-Banana Pro is that they are pretty neck & neck overall. I find GPT a lot easier to prompt, with NB, you often had to iterate several times before getting a good result, while with GPT you typically get what you ask for. But I think NB
-
Claude Opus 4.5 Shows 20% More Goal-Oriented Behavior
By
–
Claude Opus 4.5 feels 20% more goal-oriented. Two examples from today: I dictated a long client email, it drafted the email and say something like "yeah i went ahead and removed these 2 points because they don't support your point as much." I dictated train options to a
-
GPT-5 Adds Conversational Image Generation and Editing
By
–
We are adding conversational image gen and editing support to the GPT 5 family of bots later today! (2/3)
-
DTV Math Model: CPU-First Approach vs Symbolic Specialization
By
–
The DTV paper[1] from 2024 uses a 62B model specialized in math and reasoning, building up from 48.6% baseline by putting more emphasis on the symbolic side. What if we did the opposite, matching the baseline with specialized system on CPU and adding more parameters to improve?