We dive into the well defined task of automatic speech recognition (ASR), and describe why OpenAI’s Whisper, which has been integrated into ChatGPT, makes stuff up, or “hallucinates” as it’s called in the industry (bad nomenclature).
LLMS
-
One-Model-Everything Approach Reduces ASR System Reliability
By
–
Their (and Muskrat and others’) quest to build one-model-for-everything has resulted in less reliable systems than what we’ve had before, even in the well defined task of ASR. As in, historically “hallucinations” weren’t problems in ASR systems!
-
GPT-4.1 Launch: New Model for Developer Excellence
By
–
Welcome to the GPT-4.1 family! Trained with one goal: being truly great for developers. • Best-in-class at coding tasks and instruction following
• 1M token context + better comprehension
• Pushing performance across the latency curve What are your first impressions so far? -
LangMem and LangGraph Enable Type-Safe Agent Profile Learning
By
–
With LangMem & LangGraph, now it's easy to : 1) Initialize agent profiles with assistants
2) Learn and adapt these from experience All type-safe and stored in your agent's long-term memory store. -
SambaNova Llama4 Maverick Fastest LLM Cloud Performance
By
–
Find someone who looks at you like how devs look at us Experience the FASTEST #Llama4 Maverick on SambaNova Cloud
-
Advanced AI Prompts Overcome Basic Tactics Successfully
By
–
good tactic for normies but i have hit it with some big prompts and it cooked
-
Llama 4 Maverick Reaches 800 Tokens Per Second Speed
By
–
It's a bird, it's a plane, it's … an even FASTER speed for #Llama4 Maverick!
— SambaNova (@SambaNovaAI) 14 avril 2025
We're gearing up to 800 t/s! 🚀
See for yourself 👇It's a bird, it's a plane, it's … an even FASTER speed for #Llama4 Maverick! We're gearing up to 800 t/s! See for yourself
-
Perplexity’s Sonar Models Excel in Web Search
By
–
Perplexity published "web search" arena results for Sonar models. – Close top score with Gemini 2.5 Pro on search-related tasks
– Higher "Search depth" than anyone else -
Sonar Models Outperform Competitors at 50-90% Lower Cost
By
–
Even with superior performance, Sonar models remain 50-90% cheaper than leading competitors.
-
Search Arena reveals key factors in AI model preference rankings
By
–
Search Arena revealed three factors strongly correlating with human preference: • Longer responses
• Higher citation counts
• Citations from community sources Sonar models had substantially higher search depth, citing 2-3x more sources than equivalent Gemini models.