# Terminal-Bench 2.1

> Terminal-Bench 2.1 results, mostly provider-reported, stored separately from the Terminal-Bench 2.0 lane.

Canonical page: https://benchlm.ai/benchmarks/terminalbench21

- Category: [Agentic](/agentic)
- Last updated: September 25, 2026

Latest rankings: [the latest benchmark rankings](/benchmarks/terminal-bench-4)

## About Terminal-Bench 2.1

- Year: 2026
- Tasks: Terminal-based software-agent tasks
- Format: Interactive task success rate
- Difficulty: Professional software engineering
- Paper: [Terminal-Bench](https://www.tbench.ai/)

BenchLM stores Terminal-Bench 2.1 results on their own key, apart from Terminal-Bench 2.0, because the two versions are not directly comparable. Harness and effort settings vary by row.

BenchAlign v5.7 gives Terminal-Bench 2.1 8% of the Agentic reference weight, so it moves the Agentic leaderboard and the overall ranking. Reference weights are relative weights in the calibrated model, not fixed shares of a score.

## Leaderboard (61 models)

| Rank | Model | Creator | Score |
|------|-------|---------|-------|
| 1 | [SWE-2](/models/swe-2) | Cognition | 92.8% |
| 2 | [GPT-5.6 Sol](/models/gpt-5-6-sol) | OpenAI | 91.9% |
| 3 | [DeepSeek V4.1 Flash](/models/deepseek-v4-1-flash) | DeepSeek | 90.6% |
| 4 | [MiMo-V2.6-Pro](/models/mimo-v2-6-pro) | Xiaomi | 89.9% |
| 5 | [Gemini 3.8 Flash](/models/gemini-3-8-flash) | Google | 89.4% |
| 6 | [Muse Spark 1.3](/models/muse-spark-1-3) | Meta | 88.8% |
| 7 | [Kimi K3](/models/kimi-k3) | Moonshot AI | 88.3% |
| 8 | [GLM-5.3](/models/glm-5-3) | Z.AI | 88.2% |
| 9 | [Claude Mythos 5](/models/claude-mythos-5) | Anthropic | 88.0% |
| 10 | [DeepSeek V4 Pro 0813](/models/deepseek-v4-pro-0813) | DeepSeek | 87.9% |
| 11 | [MiMo-V2.6-Flash](/models/mimo-v2-6-flash) | Xiaomi | 87.6% |
| 12 | [GPT-5.6 Terra](/models/gpt-5-6-terra) | OpenAI | 87.4% |
| 13 | [Qwen3.8 Max](/models/qwen3-8-max) | Alibaba | 86.6% |
| 14 | [Ornith-1.5-397B](/models/ornith-1-5-397b) | Ornith AI | 86.1% |
| 15 | [Gemini 3.7 Flash](/models/gemini-3-7-flash) | Google | 85.8% |
| 16 | [Hy4 preview](/models/hy4-preview) | Tencent | 85.4% |
| 17 | [Step 5 Preview](/models/step-5-preview) | StepFun | 85.0% |
| 18 | [GPT-5.6 Luna](/models/gpt-5-6-luna) | OpenAI | 84.7% |
| 19 | [Claude Fable 5](/models/claude-fable) | Anthropic | 84.3% |
| 20 | [GLM-5.3-Flash](/models/glm-5-3-flash) | Z.AI | 84.3% |
| 21 | [Grok 4.5](/models/grok-4-5) | xAI | 83.3% |
| 22 | [Muse Spark 1.2](/models/muse-spark-1-2) | Meta | 82.9% |
| 23 | [DeepSeek V4 Flash 0731](/models/deepseek-v4-flash-0731) | DeepSeek | 82.7% |
| 24 | [Sakana Fugu-Ultra](/models/sakana-fugu-ultra) | Sakana AI | 82.1% |
| 25 | [SWE-1.7](/models/swe-1-7) | Cognition | 81.5% |
| 26 | [GLM-5.2](/models/glm-5-2) | Z.AI | 81.0% |
| 27 | [Claude Sonnet 5](/models/claude-sonnet-5) | Anthropic | 80.4% |
| 28 | [Sakana Fugu](/models/sakana-fugu) | Sakana AI | 80.2% |
| 29 | [Muse Spark 1.1](/models/muse-spark-1-1) | Meta | 80.0% |
| 30 | [Atria Dawn Preview](/models/atria-dawn-preview) | Shanghai Artificial Intelligence Laboratory | 78.3% |
| 31 | [Ornith-1.0-397B](/models/ornith-1-0-397b) | DeepReinforce AI | 77.5% |
| 32 | [Gemini 3.5 Flash](/models/gemini-3-5-flash) | Google | 76.2% |
| 33 | [dots3-note Preview](/models/dots3-note-preview) | Dots Studio | 75.1% |
| 34 | [Claude Opus 4.8](/models/claude-opus-4-8) | Anthropic | 74.6% |
| 35 | [Qwen3.8-27B](/models/qwen3-8-27b) | Alibaba | 73.0% |
| 36 | [Seed 2.1 Pro](/models/seed-2-1-pro) | ByteDance | 71.0% |
| 37 | [Apodex 1.1](/models/apodex-1-1) | Apodex | 70.8% |
| 38 | [Laguna S 2.1](/models/laguna-s-2-1) | Poolside | 70.2% |
| 39 | [Quasar 438B](/models/quasar-438b) | Multiverse Computing | 69.3% |
| 40 | [Ornith-1.5-35B-A3B](/models/ornith-1-5-35b-a3b) | Ornith AI | 67.8% |
| 41 | [Seed 2.1 Turbo](/models/seed-2-1-turbo) | ByteDance | 67.6% |
| 42 | [MiniMax M3](/models/minimax-m3) | MiniMax | 66.0% |
| 43 | [Pokee-Isaac 28B](/models/pokee-isaac-28b) | Pokee AI | 65.1% |
| 44 | [Inkling-Small](/models/inkling-small) | Thinking Machines Lab | 64.7% |
| 45 | [Ornith-1.0-35B](/models/ornith-1-0-35b) | DeepReinforce AI | 64.2% |
| 46 | [Inkling](/models/inkling) | Thinking Machines Lab | 63.8% |
| 47 | [MAI-Code-1.1-Flash](/models/mai-code-1-1-flash) | Microsoft | 62.9% |
| 48 | [Step 3.7 Flash](/models/step-3-7-flash) | StepFun | 59.5% |
| 49 | [Ling 3.0 Flash](/models/ling-3-0-flash) | InclusionAI | 57.0% |
| 50 | [Solar Pro 4](/models/solar-pro-4) | Upstage | 57.0% |
| 51 | [Nemotron 3 Ultra](/models/nemotron-3-ultra) | NVIDIA | 56.4% |
| 52 | [Gemini 3.5 Flash-Lite](/models/gemini-3-5-flash-lite) | Google | 54.0% |
| 53 | [Ternary Bonsai 2 27B](/models/ternary-bonsai-2-27b) | Prism ML | 52.8% |
| 54 | [Ornith-1.5-9B](/models/ornith-1-5-9b) | Ornith AI | 46.2% |
| 55 | [K-EXAONE 2.0](/models/k-exaone-2-0) | LG AI Research | 43.8% |
| 56 | [Ornith-1.0-9B](/models/ornith-1-0-9b) | DeepReinforce AI | 43.1% |
| 57 | [A.X K2](/models/a-x-k2) | SK Telecom | 36.0% |
| 58 | [Granite 4.2 30B](/models/granite-4-2-30b) | IBM | 29.2% |
| 59 | [Nemotron 3.5 Lightning 30B A3B NVFP4](/models/nemotron-3-5-lightning-30b-a3b-nvfp4) | NVIDIA | 23.5% |
| 60 | [Granite 4.2 8B](/models/granite-4-2-8b) | IBM | 20.6% |
| 61 | [MiniCPM5-2B](/models/minicpm5-2b) | OpenBMB | 8.6% |

## FAQ

### What does Terminal-Bench 2.1 measure?

Terminal-Bench 2.1 results, mostly provider-reported, stored separately from the Terminal-Bench 2.0 lane.

### Which model scores highest on Terminal-Bench 2.1?

SWE-2 by Cognition currently leads with a score of 92.8% on Terminal-Bench 2.1.

### How many models are evaluated on Terminal-Bench 2.1?

61 AI models have been evaluated on Terminal-Bench 2.1 on BenchLM.

### Does Terminal-Bench 2.1 affect BenchLM's overall score?

Yes. BenchAlign v5.7 gives Terminal-Bench 2.1 8% of the Agentic reference weight, so it moves the Agentic leaderboard and the overall ranking. Reference weights are relative weights in the calibrated model, not fixed shares of a score.

## Compare Top Models on Terminal-Bench 2.1

- [SWE-2 vs GPT-5.6 Sol](/compare/gpt-5-6-sol-vs-swe-2)
- [GPT-5.6 Sol vs DeepSeek V4.1 Flash](/compare/deepseek-v4-1-flash-vs-gpt-5-6-sol)
- [DeepSeek V4.1 Flash vs MiMo-V2.6-Pro](/compare/deepseek-v4-1-flash-vs-mimo-v2-6-pro)
- [MiMo-V2.6-Pro vs Gemini 3.8 Flash](/compare/gemini-3-8-flash-vs-mimo-v2-6-pro)