Anthropic and OpenAI Ship Cheaper Models 101 Minutes Apart, Igniting a Price War
altryne · x · 2026-09-25
One week after every lab head signed the "pace the frontier" pledge, Anthropic and OpenAI shipped big new models 101 minutes apart — and both are CHEAPER:
- Claude Opus 5.5 (Sep 22, 09:31 PT): Fable-level intelligence at 40% less than Opus 5 — input $4, cached reads cut to $0.20, output $20 (previously $5/$0.50/$25). Users report it "talks like a human again" and quotas last longer.
- GPT-6 Sol and Luna (same day, 11:12 PT): Sol priced at half of GPT-5.6; per token, Sol is half the price of Opus 5.5, and Luna is 40x cheaper again.
But raw token pricing misleads: for agentic coding, cached reads dominate the bill, so Opus 5.5's 20¢ cache pricing may win — run your own workload. Also this week: Meta puts Muse at the center of everything, Grok moves into your Tesla, and open source cloned Jev in a week.
More from Models
- Lukas Kaiser explains the recipe: distilled reasoning traces plus RL — lukaszkaiser · 2026-09-26
- LLM long arithmetic allows zero error: one wrong token fails the whole evaluation — ctjlewis · 2026-09-26
- OpenRouter: keeping up with new model launches has become a full-time job — jeff_weinstein · 2026-09-26
- Opus 5.5 writes shorter sentences and 95% fewer em dashes, but hedges rise 97% — arena · 2026-09-26
- Frontier models have 'memoized' 2x2-digit multiplication, no intermediate steps needed — ctjlewis · 2026-09-26
- DL Weekly #474: GPT-6 Sol and Luna, AI Evals guide, and an RRSI paper — dl_weekly · 2026-09-26