Claude Opus 5 is a weirder than usual release to evaluate, for two reasons.
Claude Opus 5: The System Card
Claude Opus 5 is trying to be the best of both worlds.
Opus 5 is a great model. It's not great enough.
Claude Fable 5 and Claude Mythos 5
Today we’re launching Claude Fable 5: a Mythos-class model that we’ve made safe for general use.
Introducing Claude Opus 5
Opus 5 is a step change improvement for the Opus tier powering long-running agents while delivering improvements in coding and professional work.
How good is Mythos?
Is Claude Mythos Preview a particularly big jump in AI capabilities?
Claude Opus 4.7 Is Here: Release Confirmed April 16, 2026
Claude Opus 4.7 shipped April 16, 2026. SWE-bench 87.6%, xhigh mode, /ultrareview, 2,576px vision. Read our full first-look review.
Introducing Claude Opus 4.6
Opus 4.7 isn't dumb, it's just lazy — Shimin Zhang
Some follow up experiments with Claude Opus 4.7 based on Simon Willison's Pelican Benchmark Shocker.
Prompting Claude Opus 5
Behavioral differences and prompting patterns for Claude Opus 5, covering response verbosity, agentic narration, task scoping, subagent delegation, self-correction, and output artifacts when thinking is disabled.
Claude Fable 5 and Mythos 5: The System Card
First things first: Claude Fable 5 is the new best publicly available model.
Looks not too impressive in this graphics. This is ~ on par with Opus 4.7, apparently. Opus 4.7 was >20x the cost. Full comparison: artificialanalysis.ai/models/comparisons/mimo-v2-5-… (Stated facts *not* independently verified by me! I'm not even sure I am reading that page right)
mr. TIM
Korean lab, Motif, releases a 341B model that performs on par with DSv4 (1.6T) they have some actual architectural innovations and a detailed tech report huggingface.co/Motif-Technologies/Motif-3-Be…
