Unless otherwise noted: adaptive thinking at max effort, default temperature/top_p, five-trial average; context window is evaluation-dependent and no more than 1M tokens.
未知项
每个 benchmark 的具体 context window 未逐项列明。
工具
工具条件未记录
推理预算
max
框架
未记录
协议
Terminal-Bench-Science 0.1; 70 tasks; 10 trials per task.
条件
Max effort in Claude Code --bare; no internet egress. Safeguards enabled without fallback; 16/700 trials were stopped by safeguards and failed.
Claude Haiku 5.5 · Terminal-Bench-Science · 0.1 · Scores and configurations
Claude Haiku 5.5 record for Terminal-Bench-Science · 0.1. 1 score record for this release, including 1 exact value (may include historical records). Includes 1 source location.
Claude Haiku 5.5 · Terminal-Bench-Science scores and configurations
Metric and score
Configuration and evaluation conditions
Sources
Score20.6%Reported 20.6% · Unit %
Published Unconfirmed
System Card 标准 max effort
Settings
Unless otherwise noted: adaptive thinking at max effort, default temperature/top_p, five-trial average; context window is evaluation-dependent and no more than 1M tokens.
Unknowns
每个 benchmark 的具体 context window 未逐项列明。
Tools
Tool access not recorded
Reasoning budget
max
Framework
not recorded
Protocol
Terminal-Bench-Science 0.1; 70 tasks; 10 trials per task.
Conditions
Max effort in Claude Code --bare; no internet egress. Safeguards enabled without fallback; 16/700 trials were stopped by safeguards and failed.