Model Lab Daily
Daily benchmarks, eval results, and lab notes from the people training the next generation of models.
Latest Analysis
-
Google I/O 2026 Keynote: Gemini 3.2 Flash, Android 17 Preview, Gemma 4 Unveiled
Google I/O 2026 is set for its most ambitious showcase yet, unveiling a suite of AI advancements including Gemini 3.2…
-
Meta’s Muse Spark Achieves Llama 4 Maverick-Level Capabilities at 10x Lower Cost
Meta’s Superintelligence Labs unveils Muse Spark, claiming Llama 4 Maverick-level performance at a tenth of the compute cost, reshaping AI…
-
Anthropic Secures SpaceX’s Colossus 1 Access Amid Musk’s Legal Battle with OpenAI
Anthropic’s new agreement with SpaceX for 220,000 Nvidia GPUs and 300 MW capacity at Colossus 1 reshapes the AI infrastructure…
-
OpenAI Hits $25B Revenue, Lays Groundwork For IPO Amid Microsoft Deal
OpenAI has surpassed $25 billion in annualized revenue, positioning itself for a potential IPO in late 2026, facilitated by a…
-
OpenAI Unveils GPT-5.5 as Its Most Advanced Model Yet, Expanding Cloud Access
OpenAI has launched GPT-5.5, its most advanced AI model, enhancing coding and research capabilities, while expanding access to AWS and…
-
Mira Murati’s Thinking Machines Lab Signs Multi-Billion-Dollar Google Cloud Deal
Thinking Machines Lab, led by Mira Murati, has struck a significant deal with Google Cloud, securing access to advanced Nvidia…