Image created with gemini-2.5-flash-image with claude-sonnet-4-5-20250929. Image prompt: A cinematic photograph of a majestic circular library with floor-to-ceiling shelves filled with glowing documents in deep blue and white light, golden retrieval pathways connecting them like luminous threads. At the center, an elegant two-candle birthday cake on a slate grey pedestal radiates warm red and white light, surrounded by floating index cards and data fragments, high contrast, crisp details, architectural grandeur meets celebration.
it’s quite incredible how bad Sonnet 4 is at long-context retrieval Grok-4 > GPT-5 ~ Gemini 2.5 Pro > Claude 4 Sonnet https://x.com/scaling01/status/1970661469667660100
Your RAG pipeline isn’t performing well? The problem might not be your 𝘳𝘦𝘵𝘳𝘪𝘦𝘷𝘢𝘭 or 𝘨𝘦𝘯𝘦𝘳𝘢𝘵𝘪𝘰𝘯 – it’s probably your data preparation. Most people jump straight into building RAG systems without properly preparing their data. But here’s the thing: https://x.com/femke_plantinga/status/1968691549358686357
