reasoning benchmark
Terminal-Bench 2.1 Leaderboard
Terminal-Bench 2.1 is an updated release of the Terminal-Bench benchmark that tests AI agents' ability to operate a computer via the terminal. It evaluates how well models handle real-world, end-to-end tasks autonomously, including compiling code, training models, setting up servers, system administration, data science workflows, and security tasks.
Updated Aug 17, 2026
Models28
Model coverage28
MetricScore
EvidenceB
Terminal-Bench 2.1 Ranking
Higher score ranks better on this benchmark.
Sort by
Rank | Model | Score | Percentile | Participants | Evidence | Evaluated |
|---|
| Rank01 | ModelOP | Score88.8% | Percentile100.0% | Participants28 | EvidenceC | Evaluated |
| Rank02 | ModelMA | Score88.3% | Percentile96.3% | Participants28 | EvidenceC | Evaluated |
| Rank03 | ModelZA | Score88.2% | Percentile92.6% | Participants28 | EvidenceC | Evaluated |
| Rank04 | ModelDE | Score87.9% | Percentile88.9% | Participants28 | EvidenceC | Evaluated |
| Rank05 | ModelOP | Score87.4% | Percentile85.2% | Participants28 | EvidenceC | Evaluated |
| Rank06 | ModelAC | Score86.6% | Percentile81.5% | Participants28 | EvidenceC | Evaluated |
| Rank07 | ModelGO | Score85.8% | Percentile77.8% | Participants28 | EvidenceC | Evaluated |
| Rank08 | ModelOP | Score84.7% | Percentile74.1% | Participants28 | EvidenceC | Evaluated |
| Rank09 | ModelAN | Score84.3% | Percentile70.4% | Participants28 | EvidenceC | Evaluated |
| Rank10 | ModelXA | Score83.3% | Percentile66.7% | Participants28 | EvidenceC | Evaluated |
| Rank11 | ModelME | Score82.9% | Percentile63.0% | Participants28 | EvidenceC | Evaluated |
| Rank12 | ModelDE | Score82.7% | Percentile59.3% | Participants28 | EvidenceC | Evaluated |
| Rank13 | ModelZA | Score82.7% | Percentile55.6% | Participants28 | EvidenceC | Evaluated |
| Rank14 | ModelME | Score80.0% | Percentile51.9% | Participants28 | EvidenceC | Evaluated |
| Rank15 | ModelGO | Score78.0% | Percentile48.1% | Participants28 | EvidenceC | Evaluated |
| Rank16 | ModelAC | Score73.0% | Percentile44.4% | Participants28 | EvidenceC | Evaluated |
| Rank17 | ModelTE | Score71.7% | Percentile40.7% | Participants28 | EvidenceC | Evaluated |
| Rank18 | ModelBY | Score71.0% | Percentile37.0% | Participants28 | EvidenceC | Evaluated |
| Rank19 | ModelPO | Score70.2% | Percentile33.3% | Participants28 | EvidenceC | Evaluated |
| Rank20 | ModelBY | Score67.6% | Percentile29.6% | Participants28 | EvidenceC | Evaluated |
| Rank21 | ModelMI | Score66.0% | Percentile25.9% | Participants28 | EvidenceC | Evaluated |
| Rank22 | ModelTM | Score64.7% | Percentile22.2% | Participants28 | EvidenceC | Evaluated |
| Rank23 | ModelMI | Score62.9% | Percentile18.5% | Participants28 | EvidenceC | Evaluated |
| Rank24 | ModelUP | Score57.0% | Percentile14.8% | Participants28 | EvidenceC | Evaluated |
| Rank25 | ModelNV | Score56.4% | Percentile11.1% | Participants28 | EvidenceC | Evaluated |
| Rank26 | ModelGO | Score54.0% | Percentile7.4% | Participants28 | EvidenceC | Evaluated |
| Rank27 | ModelME | Score51.7% | Percentile3.7% | Participants28 | EvidenceC | Evaluated |
| Rank28 | ModelNV | Score24.6% | Percentile0.0% | Participants28 | EvidenceC | Evaluated |
Terminal-Bench 2.1 Highlights
The leading models and scores on this benchmark.
Terminal-Bench 2.1 Score Distribution
A closer view of the leading scores on this benchmark.
The Top AI Models for Terminal-Bench 2.1
The first five results on this benchmark, with official price and output speed added where the model identity can be matched.
Ranking basisThis terminal-bench 2.1 AI model leaderboard uses descending score in the benchmark's original unit. The leaderboard ranking keeps matched price and speed data separate from benchmark evidence.
- 01OPGPT-5.6 SolOpenAI
- Score
- 88.8%
- Price
- $5.0 input / $30 output per 1M tokens
- Speed
- Up to 27 tok/s via OpenAI
Strengths
- Ranks #1 of 28 compared models
- 100th percentile on this benchmark
- C evidence result
Considerations
- This result measures Terminal-Bench 2.1, not total model capability
- 02MAKimi K3Moonshot AI
- Score
- 88.3%
- Price
- $3.0 input / $15 output per 1M tokens
- Speed
- Up to 26 tok/s via Fireworks
Strengths
- Ranks #2 of 28 compared models
- 96th percentile on this benchmark
- C evidence result
Considerations
- This result measures Terminal-Bench 2.1, not total model capability
- 03ZAGLM-5.3Zhipu AI
- Score
- 88.2%
Strengths
- Ranks #3 of 28 compared models
- 93th percentile on this benchmark
- C evidence result
Considerations
- This result measures Terminal-Bench 2.1, not total model capability
- 04DEDeepSeek-V4-Pro-0813DeepSeek
- Score
- 87.9%
- Price
- $0.43 input / $0.87 output per 1M tokens
- Speed
- Up to 33 tok/s via DeepSeek
Strengths
- Ranks #4 of 28 compared models
- 89th percentile on this benchmark
- C evidence result
Considerations
- This result measures Terminal-Bench 2.1, not total model capability
- 05OPGPT-5.6 TerraOpenAI
- Score
- 87.4%
- Price
- $2.0 input / $12 output per 1M tokens
- Speed
- Up to 35 tok/s via OpenAI
Strengths
- Ranks #5 of 28 compared models
- 85th percentile on this benchmark
- C evidence result
Considerations
- This result measures Terminal-Bench 2.1, not total model capability
Selection summary
Best AI Models for Terminal-Bench 2.1
GPT-5.6 Sol currently leads Terminal-Bench 2.1 with 88.8%. It is the top model on this specific benchmark, while the best LLM for the broader task should also be checked against other benchmarks, price and runtime.
Use this leaderboard with the supporting benchmark results and coverage details above. A leaderboard position summarizes the selected ranking signal; it does not replace workload-specific testing.
What is Terminal-Bench 2.1?
What Terminal-Bench 2.1 measures and how its scores work.
Terminal-Bench 2.1 is an updated release of the Terminal-Bench benchmark that tests AI agents' ability to operate a computer via the terminal. It evaluates how well models handle real-world, end-to-end tasks autonomously, including compiling code, training models, setting up servers, system administration, data science workflows, and security tasks.
Scores are shown in ratio. This benchmark is not independently verified and has an evidence level of B.
- Family
- Terminal-Bench 2.1
- Modality
- text
- Primary category
- reasoning
- Score direction
- higher
- LLMBoard eligible
- Yes
- Evaluation key
- terminal-bench-2.1|llm-stats-current
Benchmark scores retain their original unit. Overall score eligibility is shown separately.
FAQ
Common questions about Terminal-Bench 2.1.
Which model scores highest on Terminal-Bench 2.1?
GPT-5.6 Sol is currently ranked first with 88.8%.
What does Terminal-Bench 2.1 measure?
Terminal-Bench 2.1 is an updated release of the Terminal-Bench benchmark that tests AI agents' ability to operate a computer via the terminal. It evaluates how well models handle real-world, end-to-end tasks autonomously, including compiling code, training models, setting up servers, system administration, data science workflows, and security tasks.
Is a higher score better?
Yes. Higher values rank better for this benchmark.
How many models are compared?
28 model results are currently shown.
Does this benchmark affect the overall score?
Yes. This benchmark can contribute to the current LLMBoard capability score.