Uploaded July 2025 | Updated November 2026, 3 hours ago
Meta’s Llama 4 posted strong leaderboard scores — but when it shipped, it didn’t meet the moment. Why?
In this episode of LLMs on the Run, UC Berkeley Professor and RunLLM Co-founder Joseph Gonzalez unpacks what might’ve gone wrong with Llama 4, why Mixture-of-Experts (MoE) models are so tricky to train and release, and the evolving playbook followed by foundation model labs.
🔍 Topics in this episode:
• What a Mixture-of-Experts (MoE) model is — and why it matters
• Why releasing models on a schedule is harder than it looks
• How foundational model labs are converging on the same playbook
• Why the real race in AI isn’t just accuracy — it’s inference efficiency
📺 Watch more LLMs on the Run:
Playlist → tinyurl.com/msxmhph7
✨ Subscribe for more AI insights:
tinyurl.com/445y4sd3
🤖 Explore RunLLM, the AI Support Engineer for advanced technical support:
tinyurl.com/mr2vyffm
#LLMsOnTheRun #Llama4 #MixtureOfExperts #MetaAI #AIInfrastructure #ModelTraining #AIModelDevelopment #FoundationModels #JosephGonzalez #RunLLM
Meta’s Llama 4 posted strong leaderboard scores — but when it shipped, it didn’t meet the moment. Why?
In this episode of LLMs on the Run, UC Berkeley Professor and RunLLM Co-founder Joseph Gonzalez unpacks what might’ve gone wrong with Llama 4, why Mixture-of-Experts (MoE) models are so tricky to train and release, and the evolving playbook followed by foundation model labs.
🔍 Topics in this episode:
• What a Mixture-of-Experts (MoE) model is — and why it matters
• Why releasing models on a schedule is harder than it looks
• How foundational model labs are converging on the same playbook
• Why the real race in AI isn’t just accuracy — it’s inference efficiency
📺 Watch more LLMs on the Run:
Playlist → tinyurl.com/msxmhph7
✨ Subscribe for more AI insights:
tinyurl.com/445y4sd3
🤖 Explore RunLLM, the AI Support Engineer for advanced technical support:
tinyurl.com/mr2vyffm
#LLMsOnTheRun #Llama4 #MixtureOfExperts #MetaAI #AIInfrastructure #ModelTraining #AIModelDevelopment #FoundationModels #JosephGonzalez #RunLLM
