SWE-1.7 claims performance near GPT-4.5 and Claude Opus on coding tasks

Cognition released SWE-1.7, an AI model trained specifically for software engineering work that reportedly approaches the performance of much larger frontier models. For indie developers building with AI, this signals more capable coding assistance may become accessible without relying on pricey API calls to OpenAI or Anthropic.