Dataset Viewer
Auto-converted to Parquet Duplicate
rank_mini_14
int64
1
37
rank_full_90
int64
1
37
rank_delta
int64
-16
14
agent
stringlengths
14
87
mini_14
float64
0.06
1
full_90
float64
0.09
1
tasks_attempted
int64
85
89
1
1
0
vix (claude-opus-4-7)
1
0.9971
85
2
3
-1
logos-latest (claude-opus-4.7)
0.9846
0.8967
86
3
2
1
LemonHarness_GPT-5.3-CodeX ()
0.9231
0.8981
88
4
9
-5
pilot-real (claude-opus-4-6)
0.9143
0.8315
89
5
5
0
Forge (GPT-5.4)
0.8571
0.861
86
6
7
-1
NexAU-AHE (gpt-5.5)
0.8571
0.8517
89
7
8
-1
CodeBrain-1.5 (GPT-5.3-Codex)
0.8429
0.8494
87
8
6
2
Capy (GPT-5.5)
0.8286
0.8545
88
9
4
5
JJAgent (Multiple)
0.8135
0.8765
89
10
10
0
OB-1_GPT-5.4-GPT-5.3-Codex-Claude-Opus-4.5-Claude-Opus-4.6 ()
0.8
0.8247
89
11
17
-6
Mux (GPT-5.3-Codex)
0.8
0.7927
87
12
16
-4
Terminus-KIRA (Gemini-3.1-Pro-Preview)
0.7929
0.7973
86
13
11
2
Polaris (Claude-Opus-4.7-GPT-5.5-Gemini-3.1-Pro)
0.7857
0.8208
89
14
21
-7
Deep-Agents (GPT-5.2-Codex)
0.7857
0.7452
86
15
31
-16
hookele (gpt5.1-codex-mini)
0.7714
0.627
89
16
19
-3
Junie_CLI (Gemini-3-Flash-Preview-Gemini-3.1-Pro-Preview-Claude-Opus-4.6-GPT-5.3-Codex)
0.75
0.7545
86
17
13
4
WOZCODE (Claude-Opus-4-7)
0.7429
0.8169
89
18
24
-6
OB-1_GPT-5.3-Codex-Claude-Opus-4.5-Claude-Opus-4.6 ()
0.7429
0.7266
89
19
25
-6
Crux (Claude-Opus-4.6)
0.7202
0.7264
86
20
12
8
Meta-Harness (Claude-Opus-4.6)
0.7143
0.8194
85
21
22
-1
Droid (Claude-Opus-4.6)
0.7
0.731
87
22
18
4
Codelia (GPT-5.3-Codex)
0.6923
0.7891
87
23
20
3
CodeBrain-1 (GPT-5.3-Codex)
0.6769
0.7471
87
24
15
9
LemonCode (GPT-5.3-Codex)
0.6762
0.8045
89
25
23
2
WozCode (Claude-Opus-4.6)
0.6714
0.7272
87
26
26
0
MAYA (Claude-4.6-opus)
0.6714
0.7247
89
27
28
-1
Ante (Gemini-3-Pro-Preview)
0.6476
0.6755
87
28
14
14
Droid (GPT-5.3-Codex)
0.6462
0.8144
88
29
27
2
clnkr (GPT-5.5)
0.6429
0.7129
85
30
30
0
CodeBrain-1 (Gemini-3-Pro-Preview)
0.6167
0.6571
87
31
32
-1
grok-cli (grok-4.20-0309-reasoning)
0.6083
0.5812
86
32
29
3
OpenSage (Gemini-3-Pro-Preview)
0.5714
0.6614
88
33
33
0
Gemini_CLI (Gemini-3-Flash-Preview)
0.5154
0.5231
86
34
35
-1
ClaudeCode (GLM-4.7)
0.4615
0.3596
85
35
34
1
MAYA (Claude-4.5-sonnet)
0.4286
0.427
89
36
36
0
little-coder (qwen3.6-35b-a3b)
0.1857
0.2403
89
37
37
0
little-coder (qwen3.5-9b)
0.0643
0.091
89

terminal-bench-mini

Fourteen of Terminal-Bench 2.0's ninety tasks, picked so that ranking agents on the subset reproduces ranking them on the whole benchmark.

Running ninety tasks five times each is how the official leaderboard is built. That is out of reach if you are comparing quant variants, fine-tunes or local models on your own hardware. This subset turns a multi-day sweep into a few hours.

Same approach as deepswe-mini: take the published per-task results, rank the field on the full benchmark, then find a small subset that preserves that ranking. These fourteen tasks reproduce about 87% of the full benchmark's pairwise ordering, and 97% of it when two agents are more than ten points apart.

The tasks

task field pass rate spread across agents run-to-run noise
filter-js-from-html 11.9% 0.310 0.022
install-windows-3.11 11.9% 0.258 0.090
mteb-retrieve 41.2% 0.420 0.149
db-wal-recovery 56.8% 0.448 0.097
qemu-alpine-ssh 69.3% 0.417 0.093
build-pmars 75.2% 0.361 0.122
mcmc-sampling-stan 81.5% 0.328 0.100
rstan-to-pystan 81.7% 0.329 0.102
kv-store-grpc 86.4% 0.297 0.063
bn-fit-modify 87.6% 0.295 0.056
regex-log 90.8% 0.270 0.025
custom-memory-heap-crash 91.9% 0.273 0.000
crack-7z-hash 93.5% 0.228 0.022
vulnerable-secret 94.6% 0.226 0.000

Field pass rate is the mean over 37 agents. Spread is the standard deviation of those per-agent rates: a task with spread near zero tells you nothing about who is better. Run-to-run noise is the average standard deviation across repeated trials of the same agent on the same task.

Task definitions live in Terminal-Bench 2.0. This dataset names them, it does not redistribute them.

Method

The source is every trial published to the Terminal-Bench 2.0 leaderboard: 32,803 trials, 27,405 of them scorable, across 90 tasks and 75 agent/model submissions. Trials whose environment failed to start carry exception_info and no verifier result; they are dropped rather than counted as failures. The 37 submissions that attempted at least 85 tasks form the reference ranking.

The subset is a stratified sample. Tasks that every agent passes or every agent fails carry no ranking information and are excluded first. The rest are sorted by field pass rate into seven difficulty bands; within each band the quieter half is preferred, since a task that flips between identical runs costs every future user extra trials to average out; two tasks are drawn per band with a fixed seed (20260919).

Quality is measured as pairwise agreement: over every pair of agents, how often the subset orders them the way all ninety tasks do, broken out by how far apart the pair really is.

pair separation agreement
more than 10 points 97.3%
5 to 10 points 80.1%
2 to 5 points 73.2%
under 2 points 65.1%

Agreement also rises with how many tasks you run, which size_curve.csv records: 14 tasks average 85%, 22 average 89%, 30 average 91%, 60 average 95%. If the systems you are comparing are close, add tasks or add trials.

What it is good for

Use it to answer "is this configuration roughly competitive". The top of the table is stable — the leading agent leads on both, and every agent in the full top ten stays in the mini top twelve. Maximum rank movement is 16 places, and it happens in the crowded middle where full-benchmark scores differ by fractions of a point.

Do not use it to claim one agent beats another by two points. Five of the fourteen tasks have run-to-run noise above 0.09, so run more than one trial per task or expect a few points of movement that mean nothing.

Files

  • tasks.csv — the fourteen tasks with their field statistics
  • leaderboard.csv — all 37 agents scored on the mini set and the full 90, with both ranks and the movement between them
  • mini_trials.csv — the agent × task pass rates behind that leaderboard
  • size_curve.csv — agreement against subset size, 200 random subsets per size

Reproducing

Source data is the public harborframework/terminal-bench-2-leaderboard repository. Every trial record is a small result.json sitting next to terminal logs that make the repository 121 GB, so fetch only the records:

git clone --filter=blob:none --no-checkout \
  https://huggingface.co/datasets/harborframework/terminal-bench-2-leaderboard
cd terminal-bench-2-leaderboard
git sparse-checkout set --no-cone '/**/result.json'
git checkout

That is 348 MB and about two minutes. Each record carries verifier_result.rewards.reward. Note that two task-name conventions coexist in the repository, terminal-bench/<name> and <name>; normalise them or the same task counts twice and no two agents ever overlap.

Caveats

Agreement is measured against Terminal-Bench 2.0 as scored by these 37 submissions, all of them frontier hosted models. A small local model sits below that range, where the subset is untested — though a gap that large is the case it handles most reliably.

Downloads last month
-