Miles Deutscher(@milesdeutscher):We're entering the token-efficiency era of AI. GPT-5.6 Luna hits an intelligence score of ~51 at roughly $0.06 per task. Claude Opus 5 on Low effort scores about the same at over $0.30 per task. Same intelligence. ~5x the cost difference. Pure intelligence is no longer the strongest moat. Cost-efficiency needs to be baked in as every model is "capable enough" now.

2026.07.30 17:36

We're entering the token-efficiency era of AI. GPT-5.6 Luna hits an intelligence score of ~51 at roughly $0.06 per task. Claude Opus 5 on Low effort scores about the same at over $0.30 per task. Same intelligence. ~5x the cost difference. Pure intelligence is no longer the strongest moat. Cost-efficiency needs to be baked in as every model is "capable enough" now.

OpenAI@OpenAI

2026.07.29 21:21

After deployment, we applied GPT-5.6 Sol to advance the frontier of efficiency by making itself more efficient to run. The results: - 20% lower serving costs from production GPU kernel improvements. - 15%+ better token-generation efficiency from improved speculative decoding.