工具条件未记录 / max · 输出上限 128K tokens / Claude Code 2.1.207 · The card says reasoning_effort defaults to max and recommends keeping that default for benchmark reproduction. **Terminal-Bench 3.0**: We evaluate Terminal-Bench-3 tasks with the Claude Code 2.1.207 harness (reasoning effort=max, 400K context, and 128K maximum output), reporting avg@3 over three rollouts per task. Each rollout runs in an isolated container built from the task's official image, and is capped at 600 agent turns with a 10-hour timeout. Tool Search is disabled, and the artifacts each agent produces are scored by the task's official separate verifier.
Terminal-Bench · GLM-5.3模型卡;3.0 · Reported score · 工具条件未记录 / max · 输出上限 128K tokens / Claude Code 2.1.207 · The card says reasoning_effort defaults to max and recommends keeping that default for benchmark reproduction. **Terminal-Bench 3.0**: We evaluate Terminal-Bench-3 tasks with the Claude Code 2.1.207 harness (reasoning effort=max, 400K context, and 128K maximum output), reporting avg@3 over three rollouts per task. Each rollout runs in an isolated container built from the task's official image, and is capped at 600 agent turns with a 10-hour timeout. Tool Search is disabled, and the artifacts each agent produces are scored by the task's official separate verifier.
Regular display scope. Rankings are grouped by metric and evaluation protocol. Recorded tool access, budgets, and frameworks are shown separately. Scores remain available for reference when conditions are unspecified.
Reported score
Tool access not recorded / max · Output limit 128K tokens / Claude Code 2.1.207 · The card says reasoning_effort defaults to max and recommends keeping that default for benchmark reproduction. **Terminal-Bench 3.0**: We evaluate Terminal-Bench-3 tasks with the Claude Code 2.1.207 harness (reasoning effort=max, 400K context, and 128K maximum output), reporting avg@3 over three rollouts per task. Each rollout runs in an isolated container built from the task's official image, and is capped at 600 agent turns with a 10-hour timeout. Tool Search is disabled, and the artifacts each agent produces are scored by the task's official separate verifier.
Terminal-Bench · GLM-5.3模型卡;3.0 · Reported score · Tool access not recorded / max · Output limit 128K tokens / Claude Code 2.1.207 · The card says reasoning_effort defaults to max and recommends keeping that default for benchmark reproduction. **Terminal-Bench 3.0**: We evaluate Terminal-Bench-3 tasks with the Claude Code 2.1.207 harness (reasoning effort=max, 400K context, and 128K maximum output), reporting avg@3 over three rollouts per task. Each rollout runs in an isolated container built from the task's official image, and is capped at 600 agent turns with a 10-hour timeout. Tool Search is disabled, and the artifacts each agent produces are scored by the task's official separate verifier.