Model Lab Daily
Daily benchmarks, eval results, and lab notes from the people training the next generation of models.
Latest Analysis
-
April’s Open-Weight Models: Llama 4 Scout, Mistral Medium 3, and Codestral 2 Redefine AI Frontier
April 2026 witnessed groundbreaking open-weight AI model releases, closing the gap between open and closed models, and reshaping AI capabilities.
-
OpenAI’s GPT-5.5 ‘Spud’ Released, Rivals Anthropic in Agentic AI Performance
OpenAI’s latest AI model, GPT-5.5 ‘Spud’, launched April 23. It reclaims the lead in agentic coding, challenging Anthropic’s Claude Opus…
-
Google Cloud Next 2026: $70B Revenue, $240B Backlog, and Gemini Enterprise Platform
At Google Cloud Next 2026, Google announces impressive revenue figures and unveils the Gemini Enterprise Agent Platform, aiming to redefine…
-
Google A5X With NVIDIA Vera Rubin NVL72 Transforms AI Economics in H2 2026
Google Cloud’s A5X hypercomputer platform promises to revolutionize AI inference economics, boasting a 10x reduction in cost per token and…
-
69 Countries, 1,000+ AI Policies in Flight: The EU AI Act’s August 2 Deadline Looms
The EU AI Act’s impending August 2 deadline is pressuring global compliance strategies, with 69 countries crafting over 1,000 AI…
-
Open-Source AI’s April Surge: Four Major Releases Challenge Proprietary Models
April 2026’s unprecedented open-source AI model releases shake up the landscape, challenging proprietary models on cost and performance benchmarks.