Claude

Which Claude model should your team use, and when? Opus 5.5 covers everyday coding and knowledge work, like debugging a release or redlining a contract. Fable 5.2 is for your hardest, most open-ended projects, like large migrations or architecture planning. Sonnet 5.5 fits well-scoped tasks where speed and usage matter. Haiku 5.5 handles high-volume work and subagents through the API. Mythos 5.2 is for cybersecurity and advanced biology.

Some prompting habits that helped with earlier models slow down Claude Opus 5.5. • Drop “think hard.” Opus 5.5 decides how hard to think based on what you ask. When we removed phrases like that from our prompts, replies started sooner with no clear drop in quality. • Start at Medium effort, and move up only when it stalls. In our evaluations, Opus 5.5 at Medium matched or beat Opus 5 at High. • Attach the chart and skip typing out its numbers. Even at its lowest effort setting, Opus 5.5 read chart values more accurately than Opus 5 at its highest. All seven habits are in our new guide to using Opus 5.5 for knowledge work: https://lnkd.in/eqjqNqJa

At Asana, agents have roles, profile pages, and a small group of people who can train them.

 Arnab Bose, Chief Product Officer, shared how Asana's teams work with agents, with Claude powering document work and complex tasks. Read more: https://lnkd.in/ggGJcg2E

Try matching the Claude model to the scale of the work. • Opus 5.5 for planning and building full features • Fable 5.2 for the most ambitious, open-ended technical work • Sonnet 5.5 for everyday work like fixing a bug or drafting a doc • Haiku 5.5 for the smaller tasks you do again and again

Introducing Claude Meetings. Add Claude to your team's video call and it builds docs, whiteboards, and other artifacts with you while you're still talking. Claude joins as a visible participant, and your team can see what it's building and correct it as it takes shape. Claude also takes the notes, so everyone can stay in the conversation. After the call, the notes, action items, transcript, and anything Claude built are on the meeting's page in Claude. Claude Meetings is starting to roll out in beta on Team and Enterprise plans, starting with Google Meet. On Team plans it's available by default, and on Enterprise plans an admin turns it on for all or part of the organization. Learn more: https://lnkd.in/eR_gwh8d

Prompt caching is the first recommendation the Claude Platform team gives developers who want to bring costs down. Jake Dejno and Lucca Siaudzionis explain how it works in agentic flows, where each request builds on the last and the API caches what already ran. They also cover changing effort mid-conversation without breaking the cache, for when a flow running on high effort reaches an easier step. Changing effort mid-conversation is in beta. Full conversation: https://lnkd.in/eUxVzJuF

Mariko Mori’s exploration of technology as a means of creative expression has pushed the boundaries of art for more than three decades. For her retrospective exhibition "Mariko Mori: All That Shines," she used Claude to write code for a computer-generated animation, which was presented as a new video installation. On a seven-meter-wide screen in the final gallery, a horizon parts and opens onto a world of radiant light.

Claude Fable 5.2 made Quantum ESPRESSO’s phonon code about 36x faster on average at Discovered Materials. At Jump Trading, it proved strong in every major role within a research system. And on an I-9 compliance review at Harvey, it gave the client clearer direction than Fable 5.1.

Here’s how to choose between the available Claude models. In general, use Opus 5.5 as your daily driver for coding and knowledge work. Switch to Sonnet 5.5 if you want faster replies and more runway with your usage. For your most ambitious, open-ended technical and research work, reach for Fable 5.2. If you’re building on the API, Haiku 5.5 is our fastest and lightest model for high-volume work behind the scenes. These are rough guidelines, not rules—the best model is the one that fits the way you work.