# Terminal Agent Tasks on TerminalBench 2.0 Source: [Terminal Agent Tasks on TerminalBench 2.0]() Collection index: [llms.txt]() Data updated: 2026-06-23T11:53:02.381Z Evaluation of model performance on Terminal Agent Tasks using the TerminalBench 2.0 dataset, measured by pass@k success rates (pass@1, pass@3, pass@5). A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. | Scope | Recorded value | | --- | --- | | Task | Terminal Agent Tasks | | Dataset (including any stated split) | TerminalBench 2.0 | | Metric | Success Score | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 34 of 34 | | Coverage | All results currently returned for this benchmark | Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol. ## Metrics - [Success Score](): Higher is better; unit: PERCENT. - [Pass@1 Rate](): Higher is better; unit: PERCENT. - [Pass@3 Rate](): Higher is better; unit: PERCENT. - [Pass@5 Rate](): Higher is better; unit: PERCENT. ## Recorded results | Method | Success Score | Result date | Report paper | | --- | --- | --- | --- | | Claude-Opus-4.5 | 57.8 | Not recorded | 6a3a13ab206542251d7c77a0 | | Gemini 3 Pro Preview | 56.9 | Not recorded | 6a3a13ab206542251d7c77a0 | | GPT-5.2 | 54 | Not recorded | 6a3a13ab206542251d7c77a0 | | GLM-4.7 | 41 | Not recorded | 6a3a13ab206542251d7c77a0 | | CLI-Universe-32B | 33.4 | Not recorded | 6a3a13ab206542251d7c77a0 | | LiberCoder-235B | 31 | Not recorded | 6a3a13ab206542251d7c77a0 | | SkillSynth-32B | 29.6 | Not recorded | 6a3a13ab206542251d7c77a0 | | Minimax-M2.1 | 29.2 | Not recorded | 6a3a13ab206542251d7c77a0 | | Kimi-K2-Instruct | 27.8 | Not recorded | 6a3a13ab206542251d7c77a0 | | Nemotron-Terminal-32B | 27.4 | Not recorded | 6a3a13ab206542251d7c77a0 | | Qwen3-Coder-480B | 23.9 | Not recorded | 6a3a13ab206542251d7c77a0 | | CLI-Universe-14B | 23 | Not recorded | 6a3a13ab206542251d7c77a0 | | TerminalTraj-32B | 22 | Not recorded | 6a3a13ab206542251d7c77a0 | | Nemotron-Terminal-14B | 20.2 | Not recorded | 6a3a13ab206542251d7c77a0 | | SkillSynth-14B | 19.9 | Not recorded | 6a3a13ab206542251d7c77a0 | | LiberCoder-32B | 19.5 | Not recorded | 6a3a13ab206542251d7c77a0 | | TerminalTraj-14B | 19.1 | Not recorded | 6a3a13ab206542251d7c77a0 | | Qwen3-Coder-30B-A3B | 14.6 | Not recorded | 6a3a13ab206542251d7c77a0 | | SkillSynth-8B | 13.5 | Not recorded | 6a3a13ab206542251d7c77a0 | | Nemotron-Terminal-8B | 13 | Not recorded | 6a3a13ab206542251d7c77a0 | | CLI-Universe-8B | 10.9 | Not recorded | 6a3a13ab206542251d7c77a0 | | TerminalTraj-7B | 10.1 | Not recorded | 6a3a13ab206542251d7c77a0 | | Qwen3-14B | 4 | Not recorded | 6a3a13ab206542251d7c77a0 | | Qwen3-32B | 3.4 | Not recorded | 6a3a13ab206542251d7c77a0 | | Qwen3-8B | 2.5 | Not recorded | 6a3a13ab206542251d7c77a0 | ### Claude-Opus-4.5 Success Score: 57.8. Result ID: 6a3a7384fbc373592c383e3d. [Reporting paper]() Recorded evaluation configuration: {"Size":"-","Scaffold":"Terminus 2"}. This record does not assert independent reproduction. ### Gemini 3 Pro Preview Success Score: 56.9. Result ID: 6a3a7384fbc373592c383e3e. [Reporting paper]() Recorded evaluation configuration: {"Size":"-","Scaffold":"Terminus 2"}. This record does not assert independent reproduction. ### GPT-5.2 Success Score: 54. Result ID: 6a3a7384fbc373592c383e3f. [Reporting paper]() Recorded evaluation configuration: {"Size":"-","Scaffold":"Terminus 2"}. This record does not assert independent reproduction. ### GLM-4.7 Success Score: 41. Result ID: 6a3a7384fbc373592c383e40. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Size":"358B","Scaffold":"Terminus 2"}. This record does not assert independent reproduction. ### CLI-Universe-32B Success Score: 33.4. Result ID: 6a3a7384fbc373592c383e53. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"32B","Scaffold":"Terminus 2"}. This record does not assert independent reproduction. ### LiberCoder-235B Success Score: 31. Result ID: 6a3a7384fbc373592c383e41. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Size":"235B","Scaffold":"OpenHands"}. This record does not assert independent reproduction. ### SkillSynth-32B Success Score: 29.6. Result ID: 6a3a7384fbc373592c383e42. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"32B","Scaffold":"Terminus 2"}. This record does not assert independent reproduction. ### Minimax-M2.1 Success Score: 29.2. Result ID: 6a3a7384fbc373592c383e43. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"229B","Scaffold":"Terminus 2"}. This record does not assert independent reproduction. ### Kimi-K2-Instruct Success Score: 27.8. Result ID: 6a3a7384fbc373592c383e44. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"1T","Scaffold":"Terminus 2"}. This record does not assert independent reproduction. ### Nemotron-Terminal-32B Success Score: 27.4. Result ID: 6a3a7384fbc373592c383e45. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"32B","Scaffold":"Terminus 2"}. This record does not assert independent reproduction. ### Qwen3-Coder-480B Success Score: 23.9. Result ID: 6a3a7384fbc373592c383e46. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Size":"480B","Scaffold":"Terminus 2"}. This record does not assert independent reproduction. ### CLI-Universe-14B Success Score: 23. Result ID: 6a3a7384fbc373592c383e54. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"14B","Scaffold":"Terminus 2"}. This record does not assert independent reproduction. ### TerminalTraj-32B Success Score: 22. Result ID: 6a3a7384fbc373592c383e47. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Size":"32B","Scaffold":"Terminus 2"}. This record does not assert independent reproduction. ### Nemotron-Terminal-14B Success Score: 20.2. Result ID: 6a3a7384fbc373592c383e48. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"14B","Scaffold":"Terminus 2"}. This record does not assert independent reproduction. ### SkillSynth-14B Success Score: 19.9. Result ID: 6a3a7384fbc373592c383e49. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"14B","Scaffold":"Terminus 2"}. This record does not assert independent reproduction. ### LiberCoder-32B Success Score: 19.5. Result ID: 6a3a7384fbc373592c383e4a. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Size":"32B","Scaffold":"OpenHands"}. This record does not assert independent reproduction. ### TerminalTraj-14B Success Score: 19.1. Result ID: 6a3a7384fbc373592c383e4b. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Size":"14B","Scaffold":"Terminus 2"}. This record does not assert independent reproduction. ### Qwen3-Coder-30B-A3B Success Score: 14.6. Result ID: 6a3a7384fbc373592c383e4c. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Size":"30B","Scaffold":"Terminus 2"}. This record does not assert independent reproduction. ### SkillSynth-8B Success Score: 13.5. Result ID: 6a3a7384fbc373592c383e4d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"8B","Scaffold":"Terminus 2"}. This record does not assert independent reproduction. ### Nemotron-Terminal-8B Success Score: 13. Result ID: 6a3a7384fbc373592c383e4e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"8B","Scaffold":"Terminus 2"}. This record does not assert independent reproduction. ### CLI-Universe-8B Success Score: 10.9. Result ID: 6a3a7384fbc373592c383e55. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"8B","Scaffold":"Terminus 2"}. This record does not assert independent reproduction. ### TerminalTraj-7B Success Score: 10.1. Result ID: 6a3a7384fbc373592c383e4f. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Size":"7B","Scaffold":"Terminus 2"}. This record does not assert independent reproduction. ### Qwen3-14B Success Score: 4. Result ID: 6a3a7384fbc373592c383e50. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"14B","Scaffold":"Terminus 2"}. This record does not assert independent reproduction. ### Qwen3-32B Success Score: 3.4. Result ID: 6a3a7384fbc373592c383e51. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"32B","Scaffold":"Terminus 2"}. This record does not assert independent reproduction. ### Qwen3-8B Success Score: 2.5. Result ID: 6a3a7384fbc373592c383e52. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"8B","Scaffold":"Terminus 2"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.