Skip to content
Latent Space

[AINews] Anthropic Claude Opus 4.7 - one step better than 4.6 in every dimension

[AINews] Anthropic Claude Opus 4.7 - literally one step better than 4.6 in every dimension

Anthropic launched Claude Opus 4.7 at the same $5/$25 per million input/output tokens; the post says 4.7-low through 4.7-high each outperform the matching higher 4.6 tiers. Reported changes include a new xhigh reasoning tier, Claude Code defaulting to xhigh, an 11-point gain on SWE-Bench Pro, and image input up to 2,576 px on the long edge (~3.75 MP). Do not overread the tokenizer change: the same input can use up to 35% more tokens, but the post says total token use still falls by up to 50% from prior equivalents.

Why it matters: Anthropic's flagship-model release fits the policy's 85–94 band. HKR-H/K/R all pass because the post gives concrete pricing, benchmark, image-limit, and token-accounting changes that hit Claude users' core coding and cost concerns.

Read the original ↗Export Markdown