Anthropic launches Claude Sonnet 4.5 for longer multi-step work
Reviewed by Helen Jones
Anthropic released Claude Sonnet 4.5, calling it its strongest model yet for coding, computer use, and long-running agent-style tasks. The company also shipped product upgrades around the model, including better support for extended work in Claude apps and API features aimed at keeping agents coherent over longer sessions. In the Claude apps, code execution and file creation for spreadsheets, slides, and documents are part of the refresh. Related guidance covers Projects, memory, and prompt practices for the Claude 4-class models. Access and rate limits still differ across Free, Pro, Team, and API plans. Performance claims rest on Anthropic’s benchmarks and evaluations. Business-critical outputs still need human checks.
Before this release, many professionals treated Claude like a sharp first-draft partner that still lost the thread on longer jobs. You restated business context, audience, and quality rules mid-task, then spent the time you saved fixing drift, weak assumptions, and generic tone. What changed is the baseline expectation for multi-step work inside Claude’s own workspace features. Stronger long-task behaviour plus memory and context tools means a well-built Project brief can carry further before it needs rescue. That shifts the bottleneck from clever prompting to clearer inputs, staged review points, and rules for when a human must approve the next step.
Analysis
Treat this as a controlled test, not a full stack swap. Pick one recurring analysis or reporting workflow you already run in a Claude Project, re-run it on Sonnet 4.5 with the same brief, and insert two hard review checkpoints where you verify facts, tone, and business fit before the model continues.
Source note
Pulse published by Collab365 Spaces, reviewed by Helen Jones on . Cite as "Anthropic launches Claude Sonnet 4.5 for longer multi-step work", Collab365 Spaces. 3 sources referenced.