Anthropic's dual codenames melon and marshmallow: picking one, not launching two
eyishazyer · x · 2026-08-24
A user noticed Anthropic running two model codenames, melon and marshmallow, both repeatedly hitting max-token limits on the same problems. Commenters suggest Anthropic typically tests two checkpoints side-by-side to choose one, not launch two separate models. Thus marshmallow 'winning' may just mean it gets picked, not that a new Opus and a new Haiku are both coming.
More from Models
- OpenAI ships GPT-Live-1.1: new voices in six languages, fewer confirmations — OpenAI · 2026-10-16
- Muse Spark model appears on LMArena, testable in Battle and Agent modes — arena · 2026-10-16
- Running a multi-model UI design arena: Paseo orchestrates four design drafts per model — vista8 · 2026-10-16
- Ilya Sutskever questions Meta license clause taxing MaaS providers above $20M revenue — ilyasu · 2026-10-16
- Muse-Spark open-weights drop: 816B-A42B MoE model gets 359GB MLX 3-bit build — ns123abc · 2026-10-16
- Midas flags inconsistent Claude limits: 33% weekly and 45% window usage despite $100 credits, yet out of API credits — TheMoonMidas · 2026-10-16