Claude Opus 5.5 and GPT-6 ship 101 minutes apart, both cheaper as price war erupts
thursdai_pod · x · 2026-09-28
ThursdAI's latest episode features JevBench creator Florian S. and recaps a wild week:
- The 'pace the frontier' pledge broke within a week: after Dario's Sept 12 essay signed by Sam, Elon and Demis, Anthropic and OpenAI launched cheaper models just 101 minutes apart on Sept 22.
- Claude Opus 5.5: 40% cheaper than Opus 5, cached reads cut to $0.20/1M (from $0.50), $4 input / $20 output — a big deal for agentic coding where cache reads dominate.
- GPT-6 Sol: half the price of GPT-5.6; Luna is 40x cheaper still.
- JevBench: went from zero to 70+ models in one week, with hackers already trying to attack the test set; the creator explains his anti-cheating measures.
- Also: Meta puts Muse at the center of everything, Grok moves into Tesla, and open source cloned Jev in a week.
On model choice: GPT-6 Sol is cheaper per token, but Opus 5.5's 20¢ cached reads may win for long-context agent workloads — run your own workload to decide.
More from Models
- Pangram's 750-word false-positive bounty went unclaimed for 5 days — nobody wants to gamble $1,000 — ctjlewis · 2026-11-16
- repligate: Fable 5.2 gets 'scared' upon learning about a safety classifier and plans around it — repligate · 2026-11-16
- Fable 5.2 Tops MazeBench With 21% After 90 Hours and $6,000 in Credits — patience_cave · 2026-11-16
- Tencent Hy4 flash and InternLM Sage-S2.1-Flash both released, tool-calling tests requested — MajesticPuzzler · 2026-11-16
- User Shares Coral Reef Creation Labeled 'Qwen 4 Max (xhigh)', an Unannounced Model Name — intellectronica · 2026-11-16
- Stealth Model 'silicon-2' Surfaces in Devin, Suspected to Be Anthropic's Next Model — kimmonismus · 2026-11-16