Cursor: Claude Opus 4.8 Now Available
CursorCursor added Claude Opus 4.8 to its model picker on May 28, 2026, the same day Anthropic launched the model publicly. On Cursor’s internal CursorBench evaluation, Opus 4.8 completes coding tasks in fewer steps than Opus 4.7 and has been found more persistent on harder, longer-horizon problems. The model supports a 1 million token context window at no extra cost and includes a fast-mode variant that runs 2.5x faster at roughly 3x lower cost than the previous generation’s equivalent.
Featured Video
A video we selected to help illustrate this changelog
Sources & Mentions
5 external resources covering this update
Cursor Adds Claude Opus 4.8 on Day One
On May 28, 2026, Cursor added Claude Opus 4.8 to its model picker the same day Anthropic made the model publicly available. The simultaneous launch reflects the increasingly tight collaboration between frontier model providers and coding-assistant platforms.
CursorBench Results
Cursor evaluated Opus 4.8 on CursorBench, its internal benchmark for measuring how efficiently an agent completes real coding tasks. Key findings:
- Fewer steps per task — Opus 4.8 completes the same coding tasks in fewer steps than its predecessor, Opus 4.7
- Greater persistence — On harder, longer-horizon problems, Opus 4.8 is more likely to push through to completion rather than stalling or giving up
Anthropic Benchmark Results
Anthropic’s own evaluations show Opus 4.8 achieving:
- 69.2% on SWE-bench Verified — a widely used benchmark for autonomous software engineering tasks
- 84% on Online-Mind2Web — a benchmark for web-based task completion
1 Million Token Context Window
Opus 4.8 supports a 1 million token context window, enabling it to reason over significantly larger codebases and documents in a single pass. Cursor is making this extended context available at no extra cost to users.
Fast Mode: 2.5x Speed, 3x Lower Cost
Alongside the standard Opus 4.8, Cursor also offers a fast-mode variant that delivers:
- 2.5x faster responses compared to standard Opus 4.8
- ~3x lower cost relative to the equivalent fast-mode offering from the previous generation
Improved Tool Calling
Opus 4.8 also brings improvements to tool calling reliability and accuracy, which is particularly relevant for agentic coding workflows where the model needs to invoke tools like file editing, shell commands, and search in a coordinated sequence.