DeepSeek’s large 1.7T-parameter model, which DeepSeek calls broadly competitive with top proprietary models on agent tasks. It has a 1M tokens context window, and its weights are openly available. Through DeepSeek’s API it costs $0.66 off-peak · $1.32 peak per million input tokens and $1.98 off-peak · $3.96 peak per million output tokens.

None of the coding tools we track offer it as a built-in option. The August 13 general release replaced an April preview. None of the coding tools we track list it as a built-in option, though ClinePass and OpenCode Go include DeepSeek models.

What DeepSeek reports

  • DeepSeek reports 87.9 on Terminal-Bench 2.1
  • DeepSeek reports 62.7 on DeepSWE

Vendor-reported figures, not independently verified by Code×Radar.

Related changes

Sources

Facts last checked November 15, 2026.