Terminal Bench
ReportedTasks an agent must complete in a command-line terminal, checked by tests.
- Kind
- Published by a third party's leaderboard, collected by Epoch AI. The leaderboard is named and linked.
- Link
- https://www.tbench.ai/leaderboard/terminal-bench/2.0
- Score shown
- Epoch's column "Accuracy mean", as a percentage, higher is better; random guessing scores about 0.0%. Error bars as the source gives them.
- Results
- 202 in all: 35 for open-weight models, 163 for closed ones, 4 where Epoch does not say. 35 linked to a model page here.
Best open-weight result: zai-org/GLM-5, 52.4%, number 67 of 202 results.
| # | Model | Weights | Setting | Score | Date | Source |
|---|---|---|---|---|---|---|
| 1 | gpt-5.5_unknownOpenAI | closed | effort unknown; Agent: NexAU-AHE | 84.7%±2.1 | 23 Apr 2026 | tbench.ai |
| 2 | gpt-5.5_unknownOpenAI | closed | effort unknown; Agent: Capy | 83.1%±2.1 | 23 Apr 2026 | tbench.ai |
| 3 | gpt-5.5_unknownOpenAI | closed | effort unknown; Agent: Codex CLI | 82.2%±2.2 | 23 Apr 2026 | tbench.ai |
| 4 | gpt-5.5_unknownOpenAI | closed | effort unknown; Agent: Codex | 82.0%±2.2 | 23 Apr 2026 | tbench.ai |
| 5 | gpt-5.4-2026-03-05_unknownOpenAI | closed | effort unknown; Agent: ForgeCode | 81.8%±2.0 | 12 Mar 2026 | tbench.ai |
| 6 | claude-opus-4-7_unknownAnthropic | closed | effort unknown; Agent: WOZCODE | 80.2%±2.1 | 16 Apr 2026 | tbench.ai |
| 7 | gemini-3.1-pro-previewGoogle DeepMind | closed | Agent: TongAgents | 80.2%±2.6 | 19 Feb 2026 | tbench.ai |
| 8 | claude-opus-4-6_unknownAnthropic | closed | effort unknown; Agent: ForgeCode | 79.8%±1.6 | 12 Mar 2026 | tbench.ai |
| 9 | gpt-5.3-codexOpenAI | closed | Agent: SageAgent | 78.4%±2.2 | 5 Feb 2026 | tbench.ai |
| 10 | gemini-3.1-pro-previewGoogle DeepMind | closed | Agent: Forge Code | 78.4%±1.8 | 2 Mar 2026 | Terminal-Bench v2 Leaderboard |
| 11 | gemini-3.1-pro-previewGoogle DeepMind | closed | Agent: ForgeCode | 78.4%±1.8 | 2 Mar 2026 | tbench.ai |
| 12 | gpt-5.3-codexOpenAI | closed | Agent: Droid | 77.3%±2.2 | 5 Feb 2026 | tbench.ai |
| 13 | claude-opus-4-6_unknownAnthropic | closed | effort unknown; Agent: Meta-Harness | 76.4%±2.4 | 5 Feb 2026 | tbench.ai |
| 14 | gpt-5.3-codexOpenAI | closed | Agent: CodeBrain-1.5 | 75.8%±2.0 | 5 Feb 2026 | tbench.ai |
| 15 | gpt-5.3-codexOpenAI | closed | Agent: Codelia | 75.7%±2.2 | 5 Feb 2026 | tbench.ai |
| 16 | claude-opus-4-6_unknownAnthropic | closed | effort unknown; Agent: Capy | 75.3%±2.4 | 5 Feb 2026 | tbench.ai |
| 17 | gpt-5.3-codexOpenAI | closed | Agent: Simple Codex | 75.1%±2.4 | 5 Feb 2026 | tbench.ai |
| 18 | gemini-3.1-pro-previewGoogle DeepMind | closed | Agent: Terminus-KIRA | 74.8%±2.6 | 19 Feb 2026 | tbench.ai |
| 19 | claude-opus-4-6_unknownAnthropic | closed | effort unknown; Agent: Terminus-KIRA | 74.7%±2.6 | 5 Feb 2026 | tbench.ai |
| 20 | gpt-5.3-codexOpenAI | closed | Agent: Mux | 74.6%±2.5 | 5 Feb 2026 | tbench.ai |
| 21 | claude-opus-4-6_unknownAnthropic | closed | effort unknown; Agent: MAYA-V2 | 72.1%±2.2 | 5 Feb 2026 | tbench.ai |
| 22 | claude-opus-4-6_unknownAnthropic | closed | effort unknown; Agent: TongAgents | 71.9%±2.7 | 5 Feb 2026 | tbench.ai |
| 23 | gpt-5.3-codexOpenAI | closed | Agent: spoox-o-m | 71.5%±2.5 | 5 Feb 2026 | tbench.ai |
| 24 | gpt-5.3-codexOpenAI | closed | Agent: CodeBrain-1 | 70.3%±2.6 | 10 Feb 2026 | via Epoch AI |
| 25 | claude-opus-4-6Anthropic | closed | Agent: Droid | 69.9%±2.5 | 5 Feb 2026 | Terminal-Bench v2 Leaderboard |
| 26 | claude-opus-4-6_unknownAnthropic | closed | effort unknown; Agent: Droid | 69.9%±2.5 | 5 Feb 2026 | tbench.ai |
| 27 | gemini-3-pro-previewGoogle DeepMind | closed | Agent: Ante | 69.4%±2.1 | 18 Nov 2025 | tbench.ai |
| 28 | gpt-5.3-codexOpenAI | closed | Agent: IndusAGI Coding Agent | 69.1%±2.3 | 5 Feb 2026 | tbench.ai |
| 29 | claude-opus-4-6_unknownAnthropic | closed | effort unknown; Agent: Crux | 66.9% | 5 Feb 2026 | tbench.ai |
| 30 | claude-opus-4-6_unknownAnthropic | closed | effort unknown; Agent: Mux | 66.5%±2.5 | 5 Feb 2026 | tbench.ai |
| 31 | gpt-5.2-codexOpenAI | closed | Agent: Deep Agents | 66.5%±3.1 | 14 Jan 2026 | tbench.ai |
| 32 | gpt-5.5_unknownOpenAI | closed | effort unknown; Agent: clnkr | 66.1%±2.5 | 23 Apr 2026 | tbench.ai |
| 33 | gemini-3-pro-previewGoogle DeepMind | closed | Agent: SageAgent | 65.2%±2.1 | 18 Nov 2025 | tbench.ai |
| 34 | gpt-5.2-2025-12-11_unknownOpenAI | closed | effort unknown; Agent: Droid | 64.9%±2.8 | 11 Dec 2025 | tbench.ai |
| 35 | gpt-5.2-2025-12-11_mediumOpenAI | closed | effort medium; Agent: Droid | 64.9%±2.8 | 24 Dec 2025 | Terminal-Bench v2 Leaderboard |
| 36 | gpt-5.3-codexOpenAI | closed | Agent: Terminus 2 | 64.7%±2.7 | 5 Feb 2026 | tbench.ai |
| 37 | gemini-3-flash-previewGoogle DeepMind | closed | Agent: Junie CLI | 64.3%±2.8 | 17 Dec 2025 | tbench.ai |
| 38 | claude-opus-4-5-20251101_unknownAnthropic | closed | effort unknown; Agent: Droid | 63.1%±2.7 | 24 Nov 2025 | tbench.ai |
| 39 | claude-opus-4-5-20251101Anthropic | closed | Agent: Droid | 63.1%±2.7 | 11 Dec 2025 | Terminal-Bench v2 Leaderboard |
| 40 | claude-opus-4-6_unknownAnthropic | closed | effort unknown; Agent: Terminus 2 | 62.9%±2.7 | 5 Feb 2026 | tbench.ai |
| 41 | gpt-5.2-2025-12-11_unknownOpenAI | closed | effort unknown; Agent: Codex CLI | 62.9%±3.0 | 11 Dec 2025 | tbench.ai |
| 42 | gpt-5.2-2025-12-11_mediumOpenAI | closed | effort medium; Agent: Codex CLI | 62.9%±3.0 | 18 Dec 2025 | Terminal-Bench v2 Leaderboard |
| 43 | gemini-3-pro-previewGoogle DeepMind | closed | Agent: CodeBrain-1.5 | 62.2%±2.6 | 18 Nov 2025 | tbench.ai |
| 44 | gemini-3-pro-previewGoogle DeepMind | closed | Agent: CodeBrain-1 | 62.2%±2.6 | 5 Feb 2026 | via Epoch AI |
| 45 | gemini-3-pro-previewGoogle DeepMind | closed | Agent: II-Agent | 61.8%±2.8 | 18 Nov 2025 | tbench.ai |
| 46 | gpt-5.1-codex-miniOpenAI | closed | Agent: hookele | 61.6%±1.9 | 12 Nov 2025 | tbench.ai |
| 47 | gemini-3.1-pro-previewGoogle DeepMind | closed | Agent: Gemini CLI | 61.4%±4.1 | 19 Feb 2026 | tbench.ai |
| 48 | gemini-3-pro-previewGoogle DeepMind | closed | Agent: Droid | 61.1%±2.8 | 18 Nov 2025 | tbench.ai |
| 49 | gpt-5.2-2025-12-11_unknownOpenAI | closed | effort unknown; Agent: Mux | 60.7% | 11 Dec 2025 | tbench.ai |
| 50 | gpt-5.1-codex-maxOpenAI | closed | Agent: Codex CLI | 60.4%±2.7 | 19 Nov 2025 | tbench.ai |
| 51 | claude-opus-4-5-20251101_unknownAnthropic | closed | effort unknown; Agent: Letta Code | 59.1%±2.4 | 24 Nov 2025 | tbench.ai |
| 52 | claude-opus-4-5-20251101_128KAnthropic | closed | Agent: Letta Code | 59.1%±2.4 | 17 Dec 2025 | Terminal-Bench v2 Leaderboard |
| 53 | claude-opus-4-5-20251101_unknownAnthropic | closed | effort unknown; Agent: Mux | 58.4% | 24 Nov 2025 | tbench.ai |
| 54 | claude-opus-4-6_unknownAnthropic | closed | effort unknown; Agent: Claude Code | 58.0%±2.9 | 5 Feb 2026 | tbench.ai |
| 55 | gpt-5.1-codexOpenAI | closed | Agent: Codex CLI | 57.8%±2.9 | 16 Nov 2025 | Terminal-Bench v2 Leaderboard |
| 56 | claude-opus-4-5-20251101_unknownAnthropic | closed | effort unknown; Agent: Terminus 2 | 57.8%±2.5 | 24 Nov 2025 | tbench.ai |
| 57 | gpt-5.1-codexOpenAI | closed | Agent: Crux | 57.8%±2.9 | 12 Nov 2025 | tbench.ai |
| 58 | grok-4-20xAI | closed | Agent: Grok CLI | 57.3% | 10 Mar 2026 | tbench.ai |
| 59 | grok-4-20xAI | closed | Agent: grok-cli | 57.3% | 2 Apr 2026 | tbench.ai |
| 60 | gemini-3-pro-previewGoogle DeepMind | closed | Agent: Terminus 2 | 56.9%±2.5 | 18 Nov 2025 | tbench.ai |
| 61 | gemini-3-pro-previewGoogle DeepMind | closed | Agent: Letta Code | 56.0%±3.0 | 18 Nov 2025 | tbench.ai |
| 62 | claude-opus-4-5-20251101_unknownAnthropic | closed | effort unknown; Agent: Goose | 54.3%±2.6 | 24 Nov 2025 | tbench.ai |
| 63 | gpt-5.2-2025-12-11_unknownOpenAI | closed | effort unknown; Agent: Terminus 2 | 54.0%±2.9 | 11 Dec 2025 | tbench.ai |
| 64 | gpt-5.1-codexOpenAI | closed | Agent: Letta Code | 53.5%±2.8 | 12 Nov 2025 | tbench.ai |
| 65 | claude-sonnet-4-6_unknownAnthropic | closed | effort unknown; Agent: Simplai Agent | 53.4%±2.8 | 17 Feb 2026 | tbench.ai |
| 66 | claude-sonnet-4-6Anthropic | closed | Agent: Simplai Agent | 53.4%±2.8 | 14 May 2026 | via Epoch AI |
| 67 | zai-org/GLM-5Epoch: glm-5, Z.ai (Zhipu AI) | open | Agent: Terminus 2 | 52.4%±2.6 | 11 Feb 2026 | tbench.ai |
| 68 | claude-opus-4-5-20251101_unknownAnthropic | closed | effort unknown; Agent: Claude Code | 52.1%±2.5 | 24 Nov 2025 | tbench.ai |
| 69 | claude-opus-4-5-20251101_unknownAnthropic | closed | effort unknown; Agent: OpenHands | 51.9%±2.9 | 24 Nov 2025 | tbench.ai |
| 70 | claude-opus-4-5-20251101_unknownAnthropic | closed | effort unknown; Agent: OpenCode | 51.7% | 24 Nov 2025 | tbench.ai |
| 71 | gemini-3-flash-previewGoogle DeepMind | closed | Agent: Terminus 2 | 51.7%±3.1 | 17 Dec 2025 | tbench.ai |
| 72 | gemini-3-flash-previewGoogle DeepMind | closed | Agent: Gemini CLI | 51.0%±3.0 | 23 Dec 2025 | Terminal-Bench v2 Leaderboard |
| 73 | gpt-5-2025-08-07_unknownOpenAI | closed | effort unknown; Agent: Codex CLI | 49.6%±2.9 | 7 Aug 2025 | tbench.ai |
| 74 | gpt-5-2025-08-07_mediumOpenAI | closed | effort medium; Agent: Codex CLI | 49.6%±2.9 | 4 Nov 2025 | Terminal-Bench v2 Leaderboard |
| 75 | gpt-5.1-2025-11-13_mediumOpenAI | closed | effort medium; Agent: Terminus 2 | 47.6%±2.8 | 16 Nov 2025 | Terminal-Bench v2 Leaderboard |
| 76 | gpt-5.1-2025-11-13_mediumOpenAI | closed | effort medium; Agent: Terminus 2 | 47.6%±2.8 | 16 Nov 2025 | Terminal-Bench v2 Leaderboard |
| 77 | gpt-5.1-2025-11-13_unknownOpenAI | closed | effort unknown; Agent: Terminus 2 | 47.6%±2.8 | 12 Nov 2025 | tbench.ai |
| 78 | gemini-3-flash-previewGoogle DeepMind | closed | Agent: Gemini CLI | 47.4%±3.0 | 17 Dec 2025 | tbench.ai |
| 79 | claude-sonnet-4-5-20250929_unknownAnthropic | closed | effort unknown; Agent: CAMEL-AI | 46.5%±2.4 | 29 Sep 2025 | tbench.ai |
| 80 | MiniMaxAI/MiniMax-M2.7Epoch: MiniMax-M2.7, MiniMax | open | Agent: IndusAGI Coding Agent | 45.1% | 18 Mar 2026 | tbench.ai |
| 81 | gpt-5-codexOpenAI | closed | Agent: Codex CLI | 44.3%±2.7 | 15 Sep 2025 | tbench.ai |
| 82 | gpt-5-codexOpenAI | closed | Agent: Codex CLI | 44.3%±2.7 | 4 Nov 2025 | Terminal-Bench v2 Leaderboard |
| 83 | gpt-5-2025-08-07_unknownOpenAI | closed | effort unknown; Agent: OpenHands | 43.8%±3.0 | 7 Aug 2025 | tbench.ai |
| 84 | gpt-5-2025-08-07_mediumOpenAI | closed | effort medium; Agent: OpenHands | 43.8%±3.0 | 2 Nov 2025 | Terminal-Bench v2 Leaderboard |
| 85 | gpt-5-codexOpenAI | closed | Agent: Terminus 2 | 43.4%±2.9 | 15 Sep 2025 | tbench.ai |
| 86 | gpt-5-codexOpenAI | closed | Agent: Terminus 2 | 43.4%±2.9 | 31 Oct 2025 | Terminal-Bench v2 Leaderboard |
| 87 | moonshotai/Kimi-K2.5Epoch: kimi-k2.5, Moonshot | open | Agent: Terminus 2 | 43.2%±2.9 | 27 Jan 2026 | tbench.ai |
| 88 | claude-sonnet-4-5-20250929_unknownAnthropic | closed | effort unknown; Agent: Goose | 43.1%±2.6 | 29 Sep 2025 | tbench.ai |
| 89 | gpt-5.1-codex-miniOpenAI | closed | Agent: Crux | 43.1%±3.0 | 12 Nov 2025 | tbench.ai |
| 90 | gpt-5.1-codex-miniOpenAI | closed | Agent: Codex CLI | 43.1%±3.0 | 17 Nov 2025 | Terminal-Bench v2 Leaderboard |
| 91 | MiniMaxAI/MiniMax-M2.7Epoch: MiniMax-M2.7, MiniMax | open | Agent: Harness Agent | 42.9%±2.9 | 18 Mar 2026 | tbench.ai |
| 92 | claude-sonnet-4-5-20250929_unknownAnthropic | closed | effort unknown; Agent: Terminus 2 | 42.8%±2.8 | 29 Sep 2025 | tbench.ai |
| 93 | claude-sonnet-4-5-20250929Anthropic | closed | Agent: Terminus 2 | 42.8%±2.8 | 31 Oct 2025 | Terminal-Bench v2 Leaderboard |
| 94 | claude-sonnet-4-5-20250929_unknownAnthropic | closed | effort unknown; Agent: MAYA | 42.7% | 4 Jan 2026 | via Epoch AI |
| 95 | claude-sonnet-4-5-20250929_unknownAnthropic | closed | effort unknown; Agent: MAYA-V2 | 42.7% | 29 Sep 2025 | tbench.ai |
| 96 | MiniMaxAI/MiniMax-M2.5Epoch: MiniMax-M2.5, MiniMax | open | Agent: cchuter | 42.7%±2.8 | 12 Feb 2026 | tbench.ai |
| 97 | claude-sonnet-4-5-20250929Anthropic | closed | Agent: OpenHands | 42.6%±2.8 | 2 Nov 2025 | Terminal-Bench v2 Leaderboard |
| 98 | claude-sonnet-4-5-20250929_unknownAnthropic | closed | effort unknown; Agent: OpenHands | 42.6%±2.8 | 29 Sep 2025 | tbench.ai |
| 99 | claude-sonnet-4-5-20250929_unknownAnthropic | closed | effort unknown; Agent: Mini-SWE-Agent | 42.5%±2.8 | 29 Sep 2025 | tbench.ai |
| 100 | claude-sonnet-4-5-20250929Anthropic | closed | Agent: Mini-SWE-Agent | 42.5%±2.8 | 3 Nov 2025 | Terminal-Bench v2 Leaderboard |
| 101 | MiniMaxAI/MiniMax-M2.5Epoch: MiniMax-M2.5, MiniMax | open | Agent: Terminus 2 | 42.2%±2.6 | 12 Feb 2026 | tbench.ai |
| 102 | gpt-5-codexOpenAI | closed | Agent: Mini-SWE-Agent | 41.3%±2.8 | 15 Sep 2025 | tbench.ai |
| 103 | gpt-5-codexOpenAI | closed | Agent: Mini-SWE-Agent | 41.3%±2.8 | 3 Nov 2025 | Terminal-Bench v2 Leaderboard |
| 104 | claude-sonnet-4-5-20250929Anthropic | closed | Agent: Claude Code | 40.1%±2.9 | 4 Nov 2025 | Terminal-Bench v2 Leaderboard |
| 105 | claude-sonnet-4-5-20250929_unknownAnthropic | closed | effort unknown; Agent: Claude Code | 40.1%±2.9 | 29 Sep 2025 | tbench.ai |
| 106 | deepseek-ai/DeepSeek-V3.2Epoch: deepseek/deepseek-v3.2, DeepSeek | open | Agent: Terminus 2 | 39.6%±2.8 | 1 Dec 2025 | tbench.ai |
| 107 | claude-opus-4-1-20250805Anthropic | closed | Agent: Terminus 2 | 38.0%±2.6 | 31 Oct 2025 | Terminal-Bench v2 Leaderboard |
| 108 | claude-opus-4-1-20250805_unknownAnthropic | closed | effort unknown; Agent: Terminus 2 | 38.0%±2.6 | 5 Aug 2025 | tbench.ai |
| 109 | claude-opus-4-1-20250805Anthropic | closed | Agent: OpenHands | 36.9%±2.7 | 2 Nov 2025 | Terminal-Bench v2 Leaderboard |
| 110 | gpt-5.1-codexOpenAI | closed | Agent: Terminus 2 | 36.9%±3.2 | 17 Nov 2025 | Terminal-Bench v2 Leaderboard |
| 111 | claude-opus-4-1-20250805_unknownAnthropic | closed | effort unknown; Agent: OpenHands | 36.9%±2.7 | 5 Aug 2025 | tbench.ai |
| 112 | gpt-5.1-codexOpenAI | closed | Agent: Terminus 2 | 36.9%±3.2 | 12 Nov 2025 | tbench.ai |
| 113 | MiniMaxAI/MiniMax-M2.1Epoch: MiniMax-M2.1, MiniMax | open | Agent: Crux | 36.6%±2.9 | 23 Dec 2025 | tbench.ai |
| 114 | moonshotai/Kimi-K2-ThinkingEpoch: kimi-k2-thinking, Moonshot | open | Agent: Terminus 2 | 35.7%±2.8 | 6 Nov 2025 | tbench.ai |
| 115 | moonshotai/Kimi-K2-ThinkingEpoch: kimi-k2-thinking, Moonshot | open | Agent: Terminus 2 | 35.7%±2.8 | 11 Nov 2025 | Terminal-Bench v2 Leaderboard |
| 116 | claude-haiku-4-5-20251001_unknownAnthropic | closed | effort unknown; Agent: Goose | 35.5%±2.9 | 15 Oct 2025 | tbench.ai |
| 117 | gpt-5-2025-08-07_mediumOpenAI | closed | effort medium; Agent: Terminus 2 | 35.2%±3.1 | 31 Oct 2025 | Terminal-Bench v2 Leaderboard |
| 118 | gpt-5-2025-08-07_unknownOpenAI | closed | effort unknown; Agent: Terminus 2 | 35.2%±3.1 | 7 Aug 2025 | tbench.ai |
| 119 | claude-opus-4-1-20250805Anthropic | closed | Agent: Mini-SWE-Agent | 35.1%±2.5 | 3 Nov 2025 | Terminal-Bench v2 Leaderboard |
| 120 | claude-opus-4-1-20250805_unknownAnthropic | closed | effort unknown; Agent: Mini-SWE-Agent | 35.1%±2.5 | 5 Aug 2025 | tbench.ai |
| 121 | claude-opus-4-1-20250805_unknownAnthropic | closed | effort unknown; Agent: Claude Code | 34.8%±2.9 | 5 Aug 2025 | tbench.ai |
| 122 | gpt-5-mini-2025-08-07_unknownOpenAI | closed | effort unknown; Agent: spoox-o-m | 34.8%±2.7 | 7 Aug 2025 | tbench.ai |
| 123 | claude-opus-4-1-20250805Anthropic | closed | Agent: Claude Code | 34.8%±2.9 | 4 Nov 2025 | Terminal-Bench v2 Leaderboard |
| 124 | gpt-5-mini-2025-08-07_unknownOpenAI | closed | effort unknown; Agent: spoox-m | 34.8%±2.7 | 24 Dec 2025 | via Epoch AI |
| 125 | gpt-5-2025-08-07_unknownOpenAI | closed | effort unknown; Agent: Mini-SWE-Agent | 33.9%±2.9 | 7 Aug 2025 | tbench.ai |
| 126 | gpt-5-2025-08-07_mediumOpenAI | closed | effort medium; Agent: Mini-SWE-Agent | 33.9%±2.9 | 3 Nov 2025 | Terminal-Bench v2 Leaderboard |
| 127 | zai-org/GLM-4.7Epoch: glm-4.7, Z.ai (Zhipu AI) | open | Agent: Terminus 2 | 33.4%±2.8 | 22 Dec 2025 | tbench.ai |
| 128 | zai-org/GLM-4.7Epoch: glm-4.7, Z.ai (Zhipu AI) | open | Agent: Crux | 33.3%±2.5 | 22 Dec 2025 | tbench.ai |
| 129 | gemini-2.5-proGoogle DeepMind | closed | Agent: Terminus 2 | 32.6%±3.0 | 17 Jun 2025 | tbench.ai |
| 130 | gemini-2.5-proGoogle DeepMind | closed | Agent: Terminus 2 | 32.6%±3.0 | 31 Oct 2025 | Terminal-Bench v2 Leaderboard |
| 131 | gpt-5-mini-2025-08-07_mediumOpenAI | closed | effort medium; Agent: Codex CLI | 31.9%±3.0 | 4 Nov 2025 | Terminal-Bench v2 Leaderboard |
| 132 | gpt-5-mini-2025-08-07_unknownOpenAI | closed | effort unknown; Agent: Codex CLI | 31.9%±3.0 | 7 Aug 2025 | tbench.ai |
| 133 | MiniMaxAI/MiniMax-M2Epoch: MiniMax-M2, MiniMax | open | Agent: Terminus 2 | 30.0%±2.7 | 27 Oct 2025 | tbench.ai |
| 134 | claude-haiku-4-5-20251001_unknownAnthropic | closed | effort unknown; Agent: Mini-SWE-Agent | 29.8%±2.5 | 15 Oct 2025 | tbench.ai |
| 135 | claude-haiku-4-5-20251001Anthropic | closed | Agent: Mini-SWE-Agent | 29.8%±2.5 | 3 Nov 2025 | Terminal-Bench v2 Leaderboard |
| 136 | gpt-5-mini-2025-08-07_unknownOpenAI | closed | effort unknown; Agent: OpenHands | 29.2%±2.8 | 7 Aug 2025 | tbench.ai |
| 137 | MiniMaxAI/MiniMax-M2.1Epoch: MiniMax-M2.1, MiniMax | open | Agent: Terminus 2 | 29.2%±2.9 | 23 Dec 2025 | tbench.ai |
| 138 | gpt-5-mini-2025-08-07_mediumOpenAI | closed | effort medium; Agent: OpenHands | 29.2%±2.8 | 2 Nov 2025 | Terminal-Bench v2 Leaderboard |
| 139 | claude-haiku-4-5-20251001_unknownAnthropic | closed | effort unknown; Agent: Terminus 2 | 28.3%±2.9 | 15 Oct 2025 | tbench.ai |
| 140 | claude-haiku-4-5-20251001Anthropic | closed | Agent: Terminus 2 | 28.3%±2.9 | 31 Oct 2025 | Terminal-Bench v2 Leaderboard |
| 141 | moonshotai/Kimi-K2-InstructEpoch: Kimi-K2-Instruct, Moonshot | open | Agent: Terminus 2 | 27.8%±2.5 | 1 Nov 2025 | Terminal-Bench v2 Leaderboard |
| 142 | moonshotai/Kimi-K2-InstructEpoch: Kimi-K2-Instruct, Moonshot | open | Agent: Terminus 2 | 27.8%±2.5 | 11 Jul 2025 | tbench.ai |
| 143 | claude-haiku-4-5-20251001_unknownAnthropic | closed | effort unknown; Agent: Claude Code | 27.5%±2.8 | 15 Oct 2025 | tbench.ai |
| 144 | claude-haiku-4-5-20251001Anthropic | closed | Agent: Claude Code | 27.5%±2.8 | 4 Nov 2025 | Terminal-Bench v2 Leaderboard |
| 145 | grok-4-0709xAI | closed | Agent: OpenHands | 27.2%±3.1 | 2 Nov 2025 | Terminal-Bench v2 Leaderboard |
| 146 | grok-4-0709xAI | closed | Agent: OpenHands | 27.2%±3.1 | 9 Jul 2025 | tbench.ai |
| 147 | Qwen/Qwen3-Coder-480B-A35B-InstructEpoch: Qwen3-Coder-480B-A35B-Instruct, Alibaba | open | Agent: Dakou Agent | 27.2%±2.6 | 22 Jul 2025 | tbench.ai |
| 148 | moonshotai/Kimi-K2-InstructEpoch: Kimi-K2-Instruct, Moonshot | open | Agent: OpenHands | 26.7%±2.7 | 11 Jul 2025 | tbench.ai |
| 149 | moonshotai/Kimi-K2-InstructEpoch: Kimi-K2-Instruct, Moonshot | open | Agent: OpenHands | 26.7%±2.7 | 2 Nov 2025 | Terminal-Bench v2 Leaderboard |
| 150 | gemini-2.5-proGoogle DeepMind | closed | Agent: Mini-SWE-Agent | 26.1%±2.5 | 3 Nov 2025 | Terminal-Bench v2 Leaderboard |
| 151 | gemini-2.5-proGoogle DeepMind | closed | Agent: Mini-SWE-Agent | 26.1%±2.5 | 17 Jun 2025 | tbench.ai |
| 152 | grok-code-fast-1 | not stated | Agent: Mini-SWE-Agent | 25.8%±2.6 | 28 Aug 2025 | tbench.ai |
| 153 | grok-code-fast-1 | not stated | Agent: Mini-SWE-Agent | 25.8%±2.6 | 3 Nov 2025 | Terminal-Bench v2 Leaderboard |
| 154 | grok-4-0709xAI | closed | Agent: Mini-SWE-Agent | 25.4%±2.9 | 3 Nov 2025 | Terminal-Bench v2 Leaderboard |
| 155 | Qwen/Qwen3-Coder-480B-A35B-InstructEpoch: Qwen3-Coder-480B-A35B-Instruct, Alibaba | open | Agent: OpenHands | 25.4%±2.6 | 2 Nov 2025 | Terminal-Bench v2 Leaderboard |
| 156 | grok-4-0709xAI | closed | Agent: Mini-SWE-Agent | 25.4%±2.9 | 9 Jul 2025 | tbench.ai |
| 157 | Qwen/Qwen3-Coder-480B-A35B-InstructEpoch: Qwen3-Coder-480B-A35B-Instruct, Alibaba | open | Agent: OpenHands | 25.4%±2.6 | 22 Jul 2025 | tbench.ai |
| 158 | Qwen/Qwen3.6-35B-A3BEpoch: qwen3.6-35b-a3b, Alibaba | open | Agent: little-coder | 24.6%±3.2 | 16 Apr 2026 | tbench.ai |
| 159 | zai-org/GLM-4.6Epoch: glm-4.6, Z.ai (Zhipu AI),Tsinghua University | open | Agent: Terminus 2 | 24.5%±2.4 | 30 Sep 2025 | tbench.ai |
| 160 | zai-org/GLM-4.6Epoch: glm-4.6, Z.ai (Zhipu AI),Tsinghua University | open | Agent: Terminus 2 | 24.5%±2.4 | 1 Nov 2025 | Terminal-Bench v2 Leaderboard |
| 161 | gpt-5-mini-2025-08-07_unknownOpenAI | closed | effort unknown; Agent: Terminus 2 | 24.0%±2.5 | 7 Aug 2025 | tbench.ai |
| 162 | gpt-5-mini-2025-08-07_mediumOpenAI | closed | effort medium; Agent: Terminus 2 | 24.0%±2.5 | 31 Oct 2025 | Terminal-Bench v2 Leaderboard |
| 163 | Qwen/Qwen3-Coder-480B-A35B-InstructEpoch: Qwen3-Coder-480B-A35B-Instruct, Alibaba | open | Agent: Terminus 2 | 23.9%±2.8 | 1 Nov 2025 | Terminal-Bench v2 Leaderboard |
| 164 | Qwen/Qwen3-Coder-480B-A35B-InstructEpoch: Qwen3-Coder-480B-A35B-Instruct, Alibaba | open | Agent: Terminus 2 | 23.9%±2.8 | 22 Jul 2025 | tbench.ai |
| 165 | grok-4-0709xAI | closed | Agent: Terminus 2 | 23.1%±2.9 | 9 Jul 2025 | tbench.ai |
| 166 | grok-4-0709xAI | closed | Agent: Terminus 2 | 23.1%±2.9 | 31 Oct 2025 | Terminal-Bench v2 Leaderboard |
| 167 | gpt-5-mini-2025-08-07_unknownOpenAI | closed | effort unknown; Agent: Mini-SWE-Agent | 22.2%±2.6 | 7 Aug 2025 | tbench.ai |
| 168 | gpt-5-mini-2025-08-07_mediumOpenAI | closed | effort medium; Agent: Mini-SWE-Agent | 22.2%±2.6 | 3 Nov 2025 | Terminal-Bench v2 Leaderboard |
| 169 | gpt-5-nano-2025-08-07_unknownOpenAI | closed | effort unknown; Agent: spoox-o-m | 21.8%±2.8 | 7 Aug 2025 | tbench.ai |
| 170 | gemini-2.5-proGoogle DeepMind | closed | Agent: Gemini CLI | 19.6%±2.9 | 4 Nov 2025 | Terminal-Bench v2 Leaderboard |
| 171 | gemini-2.5-proGoogle DeepMind | closed | Agent: Gemini CLI | 19.6%±2.9 | 17 Jun 2025 | tbench.ai |
| 172 | openai/gpt-oss-120bEpoch: gpt-oss-120b, OpenAI | open | Agent: Terminus 2 | 18.7%±2.7 | 1 Nov 2025 | Terminal-Bench v2 Leaderboard |
| 173 | openai/gpt-oss-120bEpoch: gpt-oss-120b_unknown, OpenAI | open | effort unknown; Agent: Terminus 2 | 18.7%±2.7 | 5 Aug 2025 | tbench.ai |
| 174 | gemini-2.5-flash-preview-09-2025Google DeepMind | closed | Agent: Mini-SWE-Agent | 17.1%±2.5 | 3 Nov 2025 | Terminal-Bench v2 Leaderboard |
| 175 | gemini-2.5-flashGoogle DeepMind | closed | Agent: Mini-SWE-Agent | 17.1%±2.5 | 17 Jun 2025 | tbench.ai |
| 176 | gemini-2.5-flash-preview-09-2025Google DeepMind | closed | Agent: Terminus 2 | 16.9%±2.4 | 31 Oct 2025 | Terminal-Bench v2 Leaderboard |
| 177 | gemini-2.5-flashGoogle DeepMind | closed | Agent: Terminus 2 | 16.9%±2.4 | 17 Jun 2025 | tbench.ai |
| 178 | gemini-2.5-flashGoogle DeepMind | closed | Agent: OpenHands | 16.4%±2.4 | 17 Jun 2025 | tbench.ai |
| 179 | gemini-2.5-proGoogle DeepMind | closed | Agent: OpenHands | 16.4%±2.8 | 17 Jun 2025 | tbench.ai |
| 180 | gemini-2.5-flash-preview-09-2025Google DeepMind | closed | Agent: OpenHands | 16.4%±2.4 | 2 Nov 2025 | Terminal-Bench v2 Leaderboard |
| 181 | gemini-2.5-proGoogle DeepMind | closed | Agent: OpenHands | 16.4%±2.8 | 2 Nov 2025 | Terminal-Bench v2 Leaderboard |
| 182 | gemini-2.5-flashGoogle DeepMind | closed | Agent: Gemini CLI | 15.4%±2.3 | 17 Jun 2025 | tbench.ai |
| 183 | gemini-2.5-flash-preview-09-2025Google DeepMind | closed | Agent: Gemini CLI | 15.4%±2.3 | 4 Nov 2025 | Terminal-Bench v2 Leaderboard |
| 184 | openai/gpt-oss-120bEpoch: gpt-oss-120b, OpenAI | open | Agent: Mini-SWE-Agent | 14.2%±2.3 | 3 Nov 2025 | Terminal-Bench v2 Leaderboard |
| 185 | grok-code-fast-1 | not stated | Agent: Terminus 2 | 14.2%±2.5 | 31 Oct 2025 | Terminal-Bench v2 Leaderboard |
| 186 | openai/gpt-oss-120bEpoch: gpt-oss-120b_unknown, OpenAI | open | effort unknown; Agent: Mini-SWE-Agent | 14.2%±2.3 | 5 Aug 2025 | tbench.ai |
| 187 | grok-code-fast-1 | not stated | Agent: Terminus 2 | 14.2%±2.5 | 28 Aug 2025 | tbench.ai |
| 188 | claude-haiku-4-5-20251001_unknownAnthropic | closed | effort unknown; Agent: OpenHands | 13.9%±2.7 | 15 Oct 2025 | tbench.ai |
| 189 | claude-haiku-4-5-20251001Anthropic | closed | Agent: OpenHands | 13.9%±2.7 | 2 Nov 2025 | Terminal-Bench v2 Leaderboard |
| 190 | gpt-5-nano-2025-08-07_mediumOpenAI | closed | effort medium; Agent: Codex CLI | 11.5%±2.3 | 4 Nov 2025 | Terminal-Bench v2 Leaderboard |
| 191 | gpt-5-nano-2025-08-07_unknownOpenAI | closed | effort unknown; Agent: Codex CLI | 11.5%±2.3 | 7 Aug 2025 | tbench.ai |
| 192 | gpt-5-nano-2025-08-07_mediumOpenAI | closed | effort medium; Agent: OpenHands | 9.9%±2.1 | 2 Nov 2025 | Terminal-Bench v2 Leaderboard |
| 193 | gpt-5-nano-2025-08-07_unknownOpenAI | closed | effort unknown; Agent: OpenHands | 9.9%±2.1 | 7 Aug 2025 | tbench.ai |
| 194 | Qwen/Qwen3.5-9BEpoch: qwen3.5-9B, Alibaba | open | Agent: little-coder | 9.2%±2.4 | 2 Mar 2026 | tbench.ai |
| 195 | gpt-5-nano-2025-08-07_mediumOpenAI | closed | effort medium; Agent: Terminus 2 | 7.9%±1.9 | 31 Oct 2025 | Terminal-Bench v2 Leaderboard |
| 196 | gpt-5-nano-2025-08-07_unknownOpenAI | closed | effort unknown; Agent: Terminus 2 | 7.9%±1.9 | 7 Aug 2025 | tbench.ai |
| 197 | gpt-5-nano-2025-08-07_mediumOpenAI | closed | effort medium; Agent: Mini-SWE-Agent | 7.0%±1.9 | 3 Nov 2025 | Terminal-Bench v2 Leaderboard |
| 198 | gpt-5-nano-2025-08-07_unknownOpenAI | closed | effort unknown; Agent: Mini-SWE-Agent | 7.0%±1.9 | 7 Aug 2025 | tbench.ai |
| 199 | openai/gpt-oss-20bEpoch: gpt-oss-20b, OpenAI | open | Agent: Mini-SWE-Agent | 3.4%±1.4 | 3 Nov 2025 | Terminal-Bench v2 Leaderboard |
| 200 | openai/gpt-oss-20bEpoch: gpt-oss-20b_unknown, OpenAI | open | effort unknown; Agent: Mini-SWE-Agent | 3.4%±1.4 | 5 Aug 2025 | tbench.ai |
| 201 | openai/gpt-oss-20bEpoch: gpt-oss-20b, OpenAI | open | Agent: Terminus 2 | 3.1%±1.5 | 1 Nov 2025 | Terminal-Bench v2 Leaderboard |
| 202 | openai/gpt-oss-20bEpoch: gpt-oss-20b_unknown, OpenAI | open | effort unknown; Agent: Terminus 2 | 3.1%±1.5 | 5 Aug 2025 | tbench.ai |
Open-weight rows are shaded. Dates are when the result was run or evaluated, where the data says. Data from Epoch AI (CC-BY 4.0), fetched 19 Nov 2026: Epoch AI, ‘Capabilities & benchmarking’. Published online at epoch.ai. Retrieved from ‘https://epoch.ai/benchmarks’ [online resource]. Epoch AI. Method