Amazon Scraps AI Leaderboards as GPT-5.6 Rumors Surface in Codex Backend

Corporate Policy

Amazon Discontinues AI Usage Leaderboards for Employees: Amazon has scrapped its internal AI leaderboard to discourage staff from using AI tools solely to increase their usage metrics. Senior executive Dave Treadwell cited rising costs and emphasized that AI should only be utilized for meaningful purposes rather than to meet arbitrary quotas.

• Amazon scraps AI leaderboard to stop workers chasing usage scores | Senior executive Dave Treadwell tells staff ‘don’t use AI just for the sake of using AI’ as costs rise
  • https://www.ft.com/content/b1a62a7f-6df5-4c90-94ce-64ce9c9961b6?syn-25a6b1a6=1

Model Benchmarking

Qwen3.6-27B Quantization Performance Analyzed: A new benchmark report evaluates various quantization methods for the Qwen3.6-27B model to determine their impact on efficiency and output quality. The analysis utilizes metrics such as Mean KLD and Same Top P Percentage to identify the most effective balance between performance and model size.

• Qwen3.6-27B Quantization Benchmark
  • https://www.huy.rocks/everyday/05-29-2026-ai-qwen3-6-27b-quantization-benchmark

Industry Rumors

Speculation Grows Over GPT-5.6 Appearance in Backend Logs: Rumors have surfaced suggesting that a "GPT-5.6" model has been detected within the backend logs of Codex. While not officially confirmed, the discovery has led to widespread speculation about an impending model update or release from OpenAI.

• GPT-5.6 spotted in Codex