Benchmark Almanac

SuperCLUE-Terminal

MiMo-V2.6 Pro 64.7 leads since September 2026
Not in model cards

Agents complete tasks in a terminal.

SuperCLUE reports scores in a public workbook.

Made by
SuperCLUE
Kind
Terminal tasks
Level
Expert
Scored
Settled by events
Answers
Agent task
Measures
Capability
In and out
Text → Actions, Text
Interaction
Agentic
Environment
Terminal
Refresh
One fixed set
Run by
A host
In views
Agents
Strongest model tested
Kimi K3 #25 on the capability index
Last model added
September 2026
Scores refreshed
November 20th, 2026
18
Models scored on its own leaderboard

MiMo-V2.6 Pro leads with 64.7 among models released by September 2026

Best score per model from the benchmark's own leaderboard, read on November 20th, 2026. Where a model ran with several settings or harnesses, we keep its best.

The newest model with a score here came out in September 2026, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

  1. 1 MiMo-V2.6 Pro api 64.7
  2. 2 Kimi K3 max / api 60.6
  3. 3 GLM-5.3 max / api 56.6
  4. 4 Tencent Hy4 preview high / api 55.6
  5. 5 MiMo-V2.6 Flash api 53.5
  6. 6 DeepSeek V4 Pro 0813 max / api 51.5
  7. 7 DeepSeek V4.1 Flash max / api 50.5
  8. 8 Qwen 3.8 Max api 49.5
  9. 9 GLM-5.2 max / api 48.5
  10. 10 GLM-5.3-Flash max / api 48.5
  11. 11 Qwen3.8 Flash api 48.5
  12. 12 DeepSeek V4 Flash 0731 max / api 46.5
  13. 13 DeepSeek-V4-Pro max / api 39.4
  14. 14 MiniMax-M3 api 36.4
  15. 15 Seed2.1 Pro high / api 35.4
  16. 16 Tencent Hy3 high / api 34.3
  17. 17 LongCat-2.0 api 33.3
  18. 18 MiMo-V2.5-Pro api 27.3

Source: Official SuperCLUE-Terminal leaderboard

#ModelScoreSettingDateSource
1 MiMo-V2.6 Pro 64.7 api Sep 2026 Official leaderboard
2 Kimi K3 60.6 max / api Jul 2026 Official leaderboard
3 GLM-5.3 56.6 max / api Aug 2026 Official leaderboard
4 Tencent Hy4 preview 55.6 high / api Aug 2026 Official leaderboard
5 MiMo-V2.6 Flash 53.5 api Sep 2026 Official leaderboard
6 DeepSeek V4 Pro 0813 51.5 max / api Aug 2026 Official leaderboard
7 DeepSeek V4.1 Flash 50.5 max / api Sep 2026 Official leaderboard
8 Qwen 3.8 Max 49.5 api Aug 2026 Official leaderboard
9 GLM-5.2 48.5 max / api Jun 2026 Official leaderboard
10 GLM-5.3-Flash 48.5 max / api Aug 2026 Official leaderboard
11 Qwen3.8 Flash 48.5 api Aug 2026 Official leaderboard
12 DeepSeek V4 Flash 0731 46.5 max / api Jul 2026 Official leaderboard
13 DeepSeek-V4-Pro 39.4 max / api Apr 2026 Official leaderboard
14 MiniMax-M3 36.4 api Jun 2026 Official leaderboard
15 Seed2.1 Pro 35.4 high / api Jun 2026 Official leaderboard
16 Tencent Hy3 34.3 high / api Jul 2026 Official leaderboard
17 LongCat-2.0 33.3 api Jul 2026 Official leaderboard
18 MiMo-V2.5-Pro 27.3 api Apr 2026 Official leaderboard

Rankings by part

We read 2 parts from the benchmark’s own leaderboard: 平均每题交互轮数 and 总题目数量. Each ranks models on one part of the test. DeepSeek V4.1 Flash leads 平均每题交互轮数 and MiMo-V2.5-Pro leads 总题目数量. Arrows show how far a model moved from its overall rank among the models we track.

平均每题交互轮数

18 models we track have a score here.

  1. 1 DeepSeek V4.1 Flash ↑6 117.7
  2. 2 GLM-5.3-Flash ↑8 116.4
  3. 3 MiniMax-M3 ↑11 106.9
  4. 4 GLM-5.3 ↓1 103.0
  5. 5 Tencent Hy4 preview ↓1 101.4
  6. 6 DeepSeek V4 Flash 0731 ↑6 91.6
  7. 7 DeepSeek V4 Pro 0813 ↓1 84.2
  8. 8 LongCat-2.0 ↑9 82.9
  9. 9 MiMo-V2.6 Flash ↓4 76.4
  10. 10 Qwen 3.8 Max ↓2 75.0

5 models have held number one on the official leaderboard since 2023

Lab claims count from the day the card came out, independent runs from the model’s release.

Xiaomi Corp DeepSeek Z.ai Moonshot AI
Own leaderboard Not tested since Sep 2026

Source: Official SuperCLUE-Terminal leaderboard

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

Plot with other benchmarks
On the benchmark’s own leaderboard
20.0 30.0 40.0 50.0 60.0 70.0 2026 20.0 30.0 40.0 50.0 60.0 70.0 2026

Source: Official SuperCLUE-Terminal leaderboard

No lab model card reports it yet

We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.

More in Terminal agents

All 35 in Terminal agents
  • SkillsBench In use 86-87 containerized expert tasks in 11 domains, run with and without Agent Skills. Verifier tests give pass rate and skill lift. GPT-5.5 66.5%
  • Terminal-Bench Hard In use Agents complete 44 hard tasks from Terminal-Bench Core in the Terminus 2 harness. Artificial Analysis scores pass@1 averaged over 3 repeats. Gemma 4 31B IT 36%
  • Terminal-Bench 4.1 In use The Terminal-Bench 4.0 tasks with hardened verifiers: agents complete 66 hard tasks in containerized terminals across software, ML, science, security, and operations. Test scripts verify each result, scored as resolution rate. Copilot Fusion 71.2%
  • Harbor-Index In use Agents solve about 80 hard tasks drawn from 29 benchmarks run through Harbor adapters, such as HLE, GSO and SWE-bench Pro. Each task keeps its source verifier. GPT-5.5 28.1%
  • Terminal-Bench 2.1 Extended Not in model cards Agents complete long command-line tasks in containers, scored by verification tests. Mercor adds 99 new expert-written tasks to the 89 public ones and reports both scores. Muse Wonder 54%
  • Terminal-Bench Pro Not in model cards Agents solve 400 terminal tasks across eight domains. Harbor distributes the 200 public tasks. Alibaba keeps the other 200 for private evaluation. Claude Opus 4.5 51%

Data audit

8 of 9 checks pass, 1 warning. Look first at the Task access check. We don’t know whether the tasks are public. Every benchmark

Catalog record 1 warning 1 pass
  • Warning Task access We don’t know whether the tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 3 pass 2 notes
  • Pass Independent runs Scores from the official leaderboard (18 entries).
  • Pass Models scored 18 models we track.
  • Pass Newest score The newest is from September 2026, for MiMo-V2.6 Flash.
  • Note Frontier models None of the top 10 has a score. The strongest tested is Kimi K3 (#25).
  • Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
  • Pass Importer We read the board on superclueai.com: 18 rows for models we track at the last read.
  • Pass Last read November 20th, 2026.
  • Pass Read worked The last read worked.
Consistency 1 pass 1 note
  • Note Lab claims No model has both a lab number and an independent run.
  • Pass Known issues No published audit found problems.

How we check