Claude Opus 5.5 max设置导致token耗尽和稳定性问题
问题概述
用户反馈在使用 Claude Opus 5.5 的 'max' 推理设置时，模型常在 128,000 token 预算耗尽前停止推理，导致无法得到完整输出；此外，Claude 5 较 4.8 更容易在复杂任务中失去上下文，表现出不稳定的行为。
痛点标签
- max设置在128,000 token预算耗尽前未能完成任务
- 模型在复杂任务中失去上下文，表现不稳定
- 高级设置的成本效益受到质疑
待验证问题
- 是否在实际使用中可以复现 max 设置提前耗尽 token budget 的情况？
- Claude 5 在复杂任务中的稳定性是否真的劣于 4.8？
- 高级设置的成本效益是否符合用户预期？
证据
原话
This is the page for the "max" reasoning setting. The page for xhigh is https://artificialanalysis.ai/models/claude-opus-5-5-xhigh and the page for medium (the default setting) is https://artificialanalysis.ai/models/claude-opus-5-5-mediumI've failed twice to get "Generate an SVG of a pelican riding a bicycle" to work with max, because in both cases it ran out of the 128,000 token budget while it was still reasoning about the problem.
原话
5 often had a tendency to get halfway through solving a problem and then I would have to stop it in the middle, because it had lost its way and was going off on a tangent rather than dealing with the problem.
原话
Half the cost per task compared to Opus 5, comparing high effort to high effort. That's just really nice.
免责声明：本卡片由 AI 从社区讨论中提炼，属于「痛点证据」，是需求分析与验证的起点，不等同于「已验证的市场需求」。AI 存在分析局限，关键决策前请打开上方原始讨论人工复核。