June 30, 2026
New model, same comment-section chaos
Claude Sonnet 5
Claude Sonnet 5 arrives, but the crowd is already asking: cool, where’s Haiku and Fable
TLDR: Claude Sonnet 5 is Anthropic’s new lower-cost AI assistant, and the big pitch is that it acts more independently while staying safer. But the comment section instantly split into hype, price skepticism, and impatient demands for other models like Haiku and the mysterious Fable.
Anthropic just dropped Claude Sonnet 5, pitching it as a smarter, cheaper AI helper that can plan ahead, use tools, and finish bigger jobs on its own. On paper, that’s a big deal: it’s now the default option for regular Claude users, and the company says it gets surprisingly close to its pricier “Opus” sibling while costing less. It’s also being sold as safer, with fewer bad behaviors and less willingness to do shady cyber stuff. Very responsible. Very polished. Very launch-day energy.
But the real show was in the comments, where the community immediately turned this into a mix of hype, nitpicking, and full-blown wishlist chaos. One user was impressed that a “prediction market” page apparently called the launch date exactly right, giving the rollout a weird little prophecy arc. Another went straight for the jugular: “When can we get a new Haiku?” In other words, nice new toy, but some fans are already complaining that the smaller model feels old. Others zoomed in on the pricing charts and basically said: if the “extra high” setting costs nearly as much as the premium model while performing a bit worse, then what exactly is the bargain here?
And then came the classic launch-thread mood swing: one person dutifully posted the system card, while another steamrolled the entire celebration with “That’s nice, but we want Fable.” Translation: the product is here, the crowd is restless, and nobody ever wants just the thing they were given.
Key Points
- •Anthropic says Claude Sonnet 5 is its most agentic Sonnet model, with planning, tool use, and autonomous task execution capabilities.
- •The company states Sonnet 5 performs close to Opus 4.8 at lower cost and improves over Sonnet 4.6 in reasoning, coding, tool use, and knowledge work.
- •Anthropic’s evaluations report that Sonnet 5 has a lower overall rate of undesirable behaviors than Sonnet 4.6 and is safer in agentic contexts.
- •The article says Sonnet 5 was not deliberately trained on cybersecurity tasks and performs substantially worse than Opus 4.8 and Mythos 5 on dangerous cyber evaluations.
- •Sonnet 5 is now available across Claude plans and via Claude Code, Claude Platform, and the Claude API, with introductory pricing through August 31, 2026.