Benchmark Almanac

Terminal-Bench Pro

Claude Opus 4.5 51% leads since December 2025
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.

On November 19, 2026, Qwen3.5 led the public set at 51.5%. Claude Opus 4.5 led the private set at 51.0%.

Made by
Alibaba
Released
December 2025
Kind
Terminal tasks
Level
Expert
Scored
Checked automatically
Answers
Agent task
Tasks
Public, plus a private test set
Measures
Capability
Language
English
In and out
Text, Code → Code, Actions
Interaction
Agentic
Environment
Terminal
Refresh
One fixed set
Run by
A host
In views
Agents
Strongest model tested
GPT-5.2 #51 on the capability index
Last model added
February 2026
Scores refreshed
November 20th, 2026
24
Entries on its own leaderboard, 23 models we track

Claude Opus 4.5 leads with 51% among models released by February 2026 Claude Opus 4.5 leads with 51% among models released by February 2026

Best score per entry from the benchmark's own leaderboard, read on November 20th, 2026. The board reports no overall score, so the headline is the plain mean of its public and private success rates. Where a model ran with several settings or harnesses, we keep its best. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. This view shows only models we track, each at its best.

The newest model with a score here came out in February 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 Claude Opus 4.5 51%
  2. 2 Qwen3.5 397B-A17B 48.7%
  3. 3 Gemini 3 Pro 48.1%
  4. 4 GPT-5.1-Codex 47%
  5. 5 Claude Sonnet 4.5 43.8%
  6. 6 DeepSeek-V3.2-Speciale 41.3%
  7. 7 GLM-4.7 38.4%
  8. 8 GPT-5.2 37.2%
  9. 9 Qwen3-Max 36%
  10. 10 Claude Haiku 4.5 35.8%
  11. 11 GLM-4.6 35.3%
  12. 12 Seed1.8 34.9%
  13. 13 Kimi K2 Sep 2025 34.8%
  14. 14 Qwen3-Coder-Plus 34%
  15. 15 DeepSeek-V3.1 33.8%
  16. 16 GPT-5 mini 32.5%
  17. 17 Qwen3-Coder-480B-A35B 32.4%
  18. 18 ROME 31%
  19. 19 gpt-oss-120b 29.9%
  20. 20 Devstral Small 2 24.6%
  21. 21 GLM-4.5-Air 24.4%
  22. 22 Gemini 2.5 Flash Sep 2025 19.4%
  23. 23 Qwen3-Coder-30B-A3B-Instruct 18.6%
  24. 24 Kimi-Linear-48B-A3B-Instruct 0.05%

Source: Official Terminal-Bench Pro leaderboard

#EntryScoreSettingDateSource
1 Claude Opus 4.5 51% Claude-Code Nov 2025 Official leaderboard
2 Qwen3.5 397B-A17B 48.7% IFlow CLI-0.4.11 Feb 2026 Official leaderboard
3 Gemini 3 Pro 48.1% IFlow CLI-0.4.11 Nov 2025 Official leaderboard
4 GPT-5.1-Codex 47% Codex-CLI Nov 2025 Official leaderboard
5 Claude Sonnet 4.5 43.8% Claude-Code Sep 2025 Official leaderboard
6 DeepSeek-V3.2-Speciale 41.3% IFlow CLI-0.4.11 Dec 2025 Official leaderboard
7 GLM-4.7 38.4% IFlow CLI-0.4.11 Dec 2025 Official leaderboard
8 GPT-5.2 37.2% IFlow CLI-0.5.7 Dec 2025 Official leaderboard
9 Qwen3-Max 36% IFlow CLI-0.4.11 Sep 2025 Official leaderboard
10 Claude Haiku 4.5 35.8% Claude-Code Oct 2025 Official leaderboard
11 GLM-4.6 35.3% IFlow CLI-0.4.11 Sep 2025 Official leaderboard
12 Seed1.8 34.9% IFlow CLI-0.4.11 Dec 2025 Official leaderboard
13 Kimi K2 Sep 2025 34.8% IFlow CLI-0.4.11 Sep 2025 Official leaderboard
14 Qwen3-Coder-Plus 34% IFlow CLI-0.4.11 Sep 2025 Official leaderboard
15 DeepSeek-V3.1 33.8% IFlow CLI-0.4.11 Aug 2025 Official leaderboard
16 GPT-5 mini 32.5% IFlow CLI-0.4.11 Aug 2025 Official leaderboard
17 Qwen3-Coder-480B-A35B 32.4% IFlow CLI-0.4.11 Jul 2025 Official leaderboard
18 ROME 30B 31% IFlow CLI-0.4.11 Official leaderboard
19 gpt-oss-120b 29.9% IFlow CLI-0.4.11 Aug 2025 Official leaderboard
20 Devstral Small 2 24.6% IFlow CLI-0.4.11 Dec 2025 Official leaderboard
21 GLM-4.5-Air 24.4% IFlow CLI-0.4.11 Jul 2025 Official leaderboard
22 Gemini 2.5 Flash Sep 2025 19.4% IFlow CLI-0.4.11 Sep 2025 Official leaderboard
23 Qwen3-Coder-30B-A3B-Instruct 18.6% IFlow CLI-0.4.11 Jul 2025 Official leaderboard
24 Kimi-Linear-48B-A3B-Instruct 0.05% IFlow CLI-0.4.11 Oct 2025 Official leaderboard
#ModelScoreSettingDateSource
1 Claude Opus 4.5 51% Claude-Code Nov 2025 Official leaderboard
2 Qwen3.5 397B-A17B 48.7% IFlow CLI-0.4.11 Feb 2026 Official leaderboard
3 Gemini 3 Pro 48.1% IFlow CLI-0.4.11 Nov 2025 Official leaderboard
4 GPT-5.1-Codex 47% Codex-CLI Nov 2025 Official leaderboard
5 Claude Sonnet 4.5 43.8% Claude-Code Sep 2025 Official leaderboard
6 DeepSeek-V3.2-Speciale 41.3% IFlow CLI-0.4.11 Dec 2025 Official leaderboard
7 GLM-4.7 38.4% IFlow CLI-0.4.11 Dec 2025 Official leaderboard
8 GPT-5.2 37.2% IFlow CLI-0.5.7 Dec 2025 Official leaderboard
9 Qwen3-Max 36% IFlow CLI-0.4.11 Sep 2025 Official leaderboard
10 Claude Haiku 4.5 35.8% Claude-Code Oct 2025 Official leaderboard
11 GLM-4.6 35.3% IFlow CLI-0.4.11 Sep 2025 Official leaderboard
12 Seed1.8 34.9% IFlow CLI-0.4.11 Dec 2025 Official leaderboard
13 Kimi K2 Sep 2025 34.8% IFlow CLI-0.4.11 Sep 2025 Official leaderboard
14 Qwen3-Coder-Plus 34% IFlow CLI-0.4.11 Sep 2025 Official leaderboard
15 DeepSeek-V3.1 33.8% IFlow CLI-0.4.11 Aug 2025 Official leaderboard
16 GPT-5 mini 32.5% IFlow CLI-0.4.11 Aug 2025 Official leaderboard
17 Qwen3-Coder-480B-A35B 32.4% IFlow CLI-0.4.11 Jul 2025 Official leaderboard
18 gpt-oss-120b 29.9% IFlow CLI-0.4.11 Aug 2025 Official leaderboard
19 Devstral Small 2 24.6% IFlow CLI-0.4.11 Dec 2025 Official leaderboard
20 GLM-4.5-Air 24.4% IFlow CLI-0.4.11 Jul 2025 Official leaderboard
21 Gemini 2.5 Flash Sep 2025 19.4% IFlow CLI-0.4.11 Sep 2025 Official leaderboard
22 Qwen3-Coder-30B-A3B-Instruct 18.6% IFlow CLI-0.4.11 Jul 2025 Official leaderboard
23 Kimi-Linear-48B-A3B-Instruct 0.05% IFlow CLI-0.4.11 Oct 2025 Official leaderboard

Rankings by part

We read 2 parts from the benchmark’s own leaderboard: Public set and Private set. Each ranks models on one part of the test. Qwen3.5 397B-A17B leads Public set and Claude Opus 4.5 leads Private set. Arrows show how far a model moved from its overall rank among the models we track.

Public set

23 models we track have a score here.

  1. 1 Qwen3.5 397B-A17B ↑1 51.5%
  2. 2 Claude Opus 4.5 ↓1 51%
  3. 3 Claude Sonnet 4.5 ↑2 49.5%
  4. 4 Gemini 3 Pro ↓1 47.9%
  5. 5 GPT-5.1-Codex ↓1 46.5%
  6. 6 DeepSeek-V3.2-Speciale 45.2%
  7. 7 Claude Haiku 4.5 ↑3 44.5%
  8. 8 GLM-4.7 ↓1 43.1%
  9. 9 GLM-4.6 ↑2 41.5%
  10. 10 Kimi K2 Sep 2025 ↑3 40.5%

Number one over time

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

Anthropic
Own leaderboard Not tested since Feb 2026

Source: Official Terminal-Bench Pro 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 Best so far
0% 20% 40% 60% 80% 100% 2026 Near the ceiling (85%) Claude Opus 4.5 Claude Sonnet 4.5 Qwen3-Max Qwen3-Coder-480B-A35B GLM-4.5-Air 0% 20% 40% 60% 80% 100% 2026 Near the ceiling (85%) Claude Opus 4.5 Claude Sonnet 4.5

Source: Official Terminal-Bench Pro 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%
  • o11y-bench Not in model cards Agents query metrics, logs and traces in a live Grafana stack, investigate incidents and fix dashboards. Grading checks claims against facts read from the same environment. Claude Opus 5.5 93.8%

Data audit

9 of 10 checks pass, 1 warning. Look first at the Newest score check. The newest came 9 months ago, in February 2026, for Qwen3.5 397B-A17B. Every benchmark

Catalog record 2 pass
  • Pass Task access Public tasks, plus a private test set.
  • Pass Confidence Checked against a primary source.
Scores 1 warning 3 pass 1 note
  • Pass Independent runs Scores from the official leaderboard (24 entries).
  • Pass Models scored 23 models we track, plus 1 other entry on its leaderboard.
  • Warning Newest score The newest came 9 months ago, in February 2026, for Qwen3.5 397B-A17B.
  • Note Frontier models None of the top 10 has a score. The strongest tested is GPT-5.2 (#51).
  • Pass Headroom The best is 51%, under 85%, so it still separates models.
Leaderboard import 3 pass
  • Pass Importer We read the board on alibaba.github.io: 23 rows for models we track and 1 other entry 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