Unverified50% confidenceTradeoffExact time
Terminal-Bench primarily covers terminal tasks and cannot reflect model performance on dimensions such as long-context understanding, multimodality, and complex reasoning.
1
Sources
50%
Confidence
Long-term
Relevance
8/11/2026
First Seen
Sources
DeepSeek V4 Flash Terminal-Bench 2.1得分82.7%意味着什么
Related Claims
UnverifiedScripts bound to a Skill are executed via the command line; the script code itself does not enter the context window, and only the execution results consume tokens.57% similarUnverifiedIn Claude Code Skills, bound scripts are executed via the command line and the script code itself does not enter the context window—only execution results consume tokens56% similarUnverifiedThe `/loop` command is session-bound: once the terminal is closed, cancelled, or exited, all tasks immediately stop and are lost with no way to recover them.55% similarUnverifiedClaude Desktop cannot directly access and modify the project file system, cannot execute terminal commands, and cannot perform context-aware reasoning at the codebase level54% similarUnverifiedSingle-session instructions are only valid during the current session and expire upon closing, whereas global custom instructions, written once, take effect permanently and apply across projects.51% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/729508API
curl https://kongchang.com/api/v1/knowledge/claims/729508MCP
get_claim(id=729508)