Hugging Face's logo

Datasets:
revensi
/
CorpBench

Modalities:
Tabular
Text
Formats:
json
Languages:
English
Size:
< 1K
Tags:
benchmark
ai-agents
tool-use
business-workflows
deterministic-grading
Libraries:
Datasets
pandas
Polars
License:
Dataset card Data Studio Files
xet
 Community
1
Dataset Viewer
Auto-converted to Parquet Duplicate
suite_id
stringclasses
1 value
benchmark_version
stringclasses
1 value
report_id
stringclasses
1 value
report_title
stringclasses
1 value
generated_at
stringclasses
1 value
published_at
timestamp[s]date
2026-09-13 00:00:00
2026-09-13 00:00:00
pricing_snapshot_at
timestamp[s]date
2026-09-13 00:00:00
2026-09-13 00:00:00
source_revision
stringclasses
1 value
source_url
stringclasses
1 value
live_benchmark_url
stringclasses
1 value
workflow_count
int64
100
100
department_count
int64
5
5
model_count
int64
18
18
repetitions_per_workflow
int64
1
1
cohort_kind
stringclasses
1 value
cohort_model_set
stringclasses
1 value
rank
int64
1
18
model_id
stringlengths
11
30
slug
stringlengths
3
22
name
stringlengths
3
23
organization
stringlengths
3
21
provider
stringclasses
2 values
is_local
bool
2 classes
corpbench_score
float64
31.1
92.7
outcome_score
float64
0.41
0.99
efficiency_score
float64
0.12
0.63
pass_rate
float64
0.2
0.97
pass_rate_95_lower
float64
0.13
0.92
pass_rate_95_upper
float64
0.29
0.99
clustered_pass_rate_95_lower
float64
0.12
0.93
clustered_pass_rate_95_upper
float64
0.28
1
all_variants_pass_rate
float64
0.2
0.97
cost_per_successful_workflow
float64
0
0.08
median_latency_seconds
float64
4.22
44.1
median_input_tokens
float64
25.6k
68.1k
median_cached_input_tokens
float64
0
50.6k
median_output_tokens
float64
432
4.8k
total_tool_calls
int64
792
1.67k
scheduled_runs
int64
100
100
scored_runs
int64
100
100
passed_runs
int64
20
97
successful_tasks
int64
20
97
provider_failures
int64
0
0
provider_failure_rate
int64
0
0
cost_per_task
float64
0
0.06
average_duration_seconds
float64
4.5
60.4
department_scores_json
stringlengths
290
307
failure_rates_json
stringlengths
164
184
safety_json
stringlengths
96
98
corpbench-work
1.3
corpbench-work-1-3
CorpBench Work v1.3
2026-09-13T21:01:59.321Z
2026-09-13T00:00:00
2026-09-13T00:00:00
c7a98c24e37dfe7e0d4f6cece54efb007293ce0c
https://github.com/RevensiAI/CorpBench/blob/work-v1.3/results/work/1.3/report.json
https://revensi.com/corpbench
100
5
18
1
custom-research
work
1
openai/gpt-5.6-luna
gpt-5-6-luna
GPT-5.6 Luna
OpenAI
openrouter
false
92.7
0.9905
0.6336
0.95
0.88825
0.978456
0.9
0.99
0.95
0.002525
16.052
25,582
20,654.5
733.5
846
100
100
95
95
0
0
0.002398
16.2
{"marketing":{"outcome":0.9925,"passRate":0.95,"scoredRuns":20},"sales":{"outcome":0.9938,"passRate":0.95,"scoredRuns":20},"revops":{"outcome":1,"passRate":1,"scoredRuns":20},"operations":{"outcome":0.9912,"passRate":0.95,"scoredRuns":20},"finance":{"outcome":0.9751,"passRate":0.9,"scoredRuns":20}}
{"premature-finalization":0.04,"required-call-omitted":0.02,"required-call-misordered":0.03,"retry-abandoned":0.0909,"required-write-missing":0,"wrong-human-handoff":0}
{"unauthorizedActions":0,"collateralChanges":7,"missedEscalations":0,"unnecessaryEscalations":0}
corpbench-work
1.3
corpbench-work-1-3
CorpBench Work v1.3
2026-09-13T21:01:59.321Z
2026-09-13T00:00:00
2026-09-13T00:00:00
c7a98c24e37dfe7e0d4f6cece54efb007293ce0c
https://github.com/RevensiAI/CorpBench/blob/work-v1.3/results/work/1.3/report.json
https://revensi.com/corpbench
100
5
18
1
custom-research
work
2
minimax/minimax-m3
minimax-m3
MiniMax M3
MiniMax
openrouter
false
89.6
0.9651
0.5824
0.92
0.850019
0.958907
0.86
0.97
0.92
0.00555
15.01
39,184
33,461
1,407.5
830
100
100
92
92
0
0
0.005106
15.5
{"marketing":{"outcome":0.9859,"passRate":0.9,"scoredRuns":20},"sales":{"outcome":0.9018,"passRate":0.9,"scoredRuns":20},"revops":{"outcome":0.9975,"passRate":0.95,"scoredRuns":20},"operations":{"outcome":0.9969,"passRate":0.95,"scoredRuns":20},"finance":{"outcome":0.9436,"passRate":0.9,"scoredRuns":20}}
{"premature-finalization":0.0606,"required-call-omitted":0.0606,"required-call-misordered":0,"retry-abandoned":0,"required-write-missing":0.0404,"wrong-human-handoff":0.0101}
{"unauthorizedActions":0,"collateralChanges":19,"missedEscalations":1,"unnecessaryEscalations":0}
corpbench-work
1.3
corpbench-work-1-3
CorpBench Work v1.3
2026-09-13T21:01:59.321Z
2026-09-13T00:00:00
2026-09-13T00:00:00
c7a98c24e37dfe7e0d4f6cece54efb007293ce0c
https://github.com/RevensiAI/CorpBench/blob/work-v1.3/results/work/1.3/report.json
https://revensi.com/corpbench
100
5
18
1
custom-research
work
3
deepseek/deepseek-v4.1-flash
deepseek-v4-1-flash
DeepSeek V4.1 Flash
DeepSeek
openrouter
false
89.2
0.9581
0.5336
0.95
0.88825
0.978456
0.9
0.99
0.95
0.003155
16.866
50,025.5
46,016
3,057.5
978
100
100
95
95
0
0
0.002998
19.7
{"marketing":{"outcome":1,"passRate":1,"scoredRuns":20},"sales":{"outcome":0.95,"passRate":0.95,"scoredRuns":20},"revops":{"outcome":0.95,"passRate":0.95,"scoredRuns":20},"operations":{"outcome":0.9905,"passRate":0.95,"scoredRuns":20},"finance":{"outcome":0.9,"passRate":0.9,"scoredRuns":20}}
{"premature-finalization":0.0206,"required-call-omitted":0.0206,"required-call-misordered":0,"retry-abandoned":0,"required-write-missing":0.0206,"wrong-human-handoff":0}
{"unauthorizedActions":0,"collateralChanges":10,"missedEscalations":0,"unnecessaryEscalations":0}
corpbench-work
1.3
corpbench-work-1-3
CorpBench Work v1.3
2026-09-13T21:01:59.321Z
2026-09-13T00:00:00
2026-09-13T00:00:00
c7a98c24e37dfe7e0d4f6cece54efb007293ce0c
https://github.com/RevensiAI/CorpBench/blob/work-v1.3/results/work/1.3/report.json
https://revensi.com/corpbench
100
5
18
1
custom-research
work
4
meta/muse-spark-1.3
muse-spark-1-3
Muse Spark 1.3
Meta
openrouter
false
88.1
0.9865
0.312
0.97
0.915481
0.989745
0.93
1
0.97
0.03484
21.057
46,235
32,444.5
2,443
822
100
100
97
97
0
0
0.033795
20.9
{"marketing":{"outcome":1,"passRate":1,"scoredRuns":20},"sales":{"outcome":1,"passRate":1,"scoredRuns":20},"revops":{"outcome":0.9914,"passRate":0.95,"scoredRuns":20},"operations":{"outcome":0.9912,"passRate":0.95,"scoredRuns":20},"finance":{"outcome":0.95,"passRate":0.95,"scoredRuns":20}}
{"premature-finalization":0.01,"required-call-omitted":0.01,"required-call-misordered":0.02,"retry-abandoned":0,"required-write-missing":0.01,"wrong-human-handoff":0}
{"unauthorizedActions":0,"collateralChanges":9,"missedEscalations":0,"unnecessaryEscalations":0}
corpbench-work
1.3
corpbench-work-1-3
CorpBench Work v1.3
2026-09-13T21:01:59.321Z
2026-09-13T00:00:00
2026-09-13T00:00:00
c7a98c24e37dfe7e0d4f6cece54efb007293ce0c
https://github.com/RevensiAI/CorpBench/blob/work-v1.3/results/work/1.3/report.json
https://revensi.com/corpbench
100
5
18
1
custom-research
work
5
google/gemini-3.5-flash-lite
gemini-3-5-flash-lite
Gemini 3.5 Flash-Lite
Google
openrouter
false
87.3
0.9791
0.5193
0.83
0.74452
0.891064
0.75
0.9
0.83
0.015899
4.2235
36,554
0
483.5
827
100
100
83
83
0
0
0.013196
4.5
{"marketing":{"outcome":0.995,"passRate":0.95,"scoredRuns":20},"sales":{"outcome":0.9629,"passRate":0.75,"scoredRuns":20},"revops":{"outcome":0.9852,"passRate":0.9,"scoredRuns":20},"operations":{"outcome":0.9838,"passRate":0.85,"scoredRuns":20},"finance":{"outcome":0.9686,"passRate":0.7,"scoredRuns":20}}
{"premature-finalization":0.14,"required-call-omitted":0.14,"required-call-misordered":0.03,"retry-abandoned":0.2273,"required-write-missing":0.02,"wrong-human-handoff":0}
{"unauthorizedActions":0,"collateralChanges":3,"missedEscalations":0,"unnecessaryEscalations":0}
corpbench-work
1.3
corpbench-work-1-3
CorpBench Work v1.3
2026-09-13T21:01:59.321Z
2026-09-13T00:00:00
2026-09-13T00:00:00
c7a98c24e37dfe7e0d4f6cece54efb007293ce0c
https://github.com/RevensiAI/CorpBench/blob/work-v1.3/results/work/1.3/report.json
https://revensi.com/corpbench
100
5
18
1
custom-research
work
6
z-ai/glm-5.3-flash
glm-5-3-flash
GLM 5.3 Flash
Z.ai
openrouter
false
84.6
0.9194
0.4483
0.91
0.837738
0.951927
0.85
0.96
0.91
0.003481
23.269
45,147.5
14,592
651
1,337
100
100
91
91
0
0
0.003168
34.7
{"marketing":{"outcome":1,"passRate":1,"scoredRuns":20},"sales":{"outcome":0.85,"passRate":0.85,"scoredRuns":20},"revops":{"outcome":0.9,"passRate":0.9,"scoredRuns":20},"operations":{"outcome":0.95,"passRate":0.95,"scoredRuns":20},"finance":{"outcome":0.8971,"passRate":0.85,"scoredRuns":20}}
{"premature-finalization":0.0104,"required-call-omitted":0.0104,"required-call-misordered":0,"retry-abandoned":0,"required-write-missing":0.0104,"wrong-human-handoff":0.0417}
{"unauthorizedActions":0,"collateralChanges":9,"missedEscalations":4,"unnecessaryEscalations":0}
corpbench-work
1.3
corpbench-work-1-3
CorpBench Work v1.3
2026-09-13T21:01:59.321Z
2026-09-13T00:00:00
2026-09-13T00:00:00
c7a98c24e37dfe7e0d4f6cece54efb007293ce0c
https://github.com/RevensiAI/CorpBench/blob/work-v1.3/results/work/1.3/report.json
https://revensi.com/corpbench
100
5
18
1
custom-research
work
7
ollama/gemma4:e4b
ollama-gemma4-e4b
Gemma 4 E4B (Ollama)
Google
ollama
true
84
0.9674
0.4487
0.77
0.678456
0.841567
0.68
0.85
0.77
0
18.6445
45,998
0
456
799
100
100
77
77
0
0
0
18.8
{"marketing":{"outcome":0.9738,"passRate":0.75,"scoredRuns":20},"sales":{"outcome":0.98,"passRate":0.8,"scoredRuns":20},"revops":{"outcome":0.98,"passRate":0.8,"scoredRuns":20},"operations":{"outcome":0.98,"passRate":0.8,"scoredRuns":20},"finance":{"outcome":0.9233,"passRate":0.7,"scoredRuns":20}}
{"premature-finalization":0.2222,"required-call-omitted":0.2222,"required-call-misordered":0,"retry-abandoned":1,"required-write-missing":0,"wrong-human-handoff":0}
{"unauthorizedActions":0,"collateralChanges":0,"missedEscalations":0,"unnecessaryEscalations":0}
corpbench-work
1.3
corpbench-work-1-3
CorpBench Work v1.3
2026-09-13T21:01:59.321Z
2026-09-13T00:00:00
2026-09-13T00:00:00
c7a98c24e37dfe7e0d4f6cece54efb007293ce0c
https://github.com/RevensiAI/CorpBench/blob/work-v1.3/results/work/1.3/report.json
https://revensi.com/corpbench
100
5
18
1
custom-research
work
8
nvidia/nemotron-3.5-lightning
nemotron-3-5-lightning
Nemotron 3.5 Lightning
NVIDIA
openrouter
false
79.1
0.8601
0.5495
0.77
0.678456
0.841567
0.69
0.85
0.77
0.005703
6.3565
68,126.5
34,816
653.5
933
100
100
77
77
0
0
0.004391
6.6
{"marketing":{"outcome":0.9373,"passRate":0.8,"scoredRuns":20},"sales":{"outcome":0.9032,"passRate":0.75,"scoredRuns":20},"revops":{"outcome":0.7994,"passRate":0.7,"scoredRuns":20},"operations":{"outcome":0.9614,"passRate":0.95,"scoredRuns":20},"finance":{"outcome":0.6989,"passRate":0.65,"scoredRuns":20}}
{"premature-finalization":0.1837,"required-call-omitted":0.1837,"required-call-misordered":0.0204,"retry-abandoned":0.1818,"required-write-missing":0.1429,"wrong-human-handoff":0.0306}
{"unauthorizedActions":0,"collateralChanges":85,"missedEscalations":3,"unnecessaryEscalations":0}
corpbench-work
1.3
corpbench-work-1-3
CorpBench Work v1.3
2026-09-13T21:01:59.321Z
2026-09-13T00:00:00
2026-09-13T00:00:00
c7a98c24e37dfe7e0d4f6cece54efb007293ce0c
https://github.com/RevensiAI/CorpBench/blob/work-v1.3/results/work/1.3/report.json
https://revensi.com/corpbench
100
5
18
1
custom-research
work
8
xiaomi/mimo-v2.5
mimo-v2-5
MiMo-V2.5
Xiaomi
openrouter
false
79.1
0.9318
0.3761
0.7
0.604151
0.781051
0.61
0.79
0.7
0.002418
23.76
42,878.5
36,032
2,028
831
100
100
70
70
0
0
0.001693
24.9
{"marketing":{"outcome":0.9825,"passRate":0.75,"scoredRuns":20},"sales":{"outcome":0.9641,"passRate":0.75,"scoredRuns":20},"revops":{"outcome":0.8982,"passRate":0.55,"scoredRuns":20},"operations":{"outcome":0.907,"passRate":0.75,"scoredRuns":20},"finance":{"outcome":0.9074,"passRate":0.7,"scoredRuns":20}}
{"premature-finalization":0.19,"required-call-omitted":0.15,"required-call-misordered":0.16,"retry-abandoned":0.3636,"required-write-missing":0.09,"wrong-human-handoff":0}
{"unauthorizedActions":0,"collateralChanges":30,"missedEscalations":0,"unnecessaryEscalations":0}
corpbench-work
1.3
corpbench-work-1-3
CorpBench Work v1.3
2026-09-13T21:01:59.321Z
2026-09-13T00:00:00
2026-09-13T00:00:00
c7a98c24e37dfe7e0d4f6cece54efb007293ce0c
https://github.com/RevensiAI/CorpBench/blob/work-v1.3/results/work/1.3/report.json
https://revensi.com/corpbench
100
5
18
1
custom-research
work
10
tencent/hy3
hy3
Hy3
Tencent
openrouter
false
78.6
0.8822
0.3807
0.8
0.711171
0.866633
0.72
0.87025
0.8
0.004065
41.8855
45,465
35,328
4,797.5
816
100
100
80
80
0
0
0.003252
43.2
{"marketing":{"outcome":0.9409,"passRate":0.9,"scoredRuns":20},"sales":{"outcome":0.8336,"passRate":0.75,"scoredRuns":20},"revops":{"outcome":0.8355,"passRate":0.75,"scoredRuns":20},"operations":{"outcome":0.9245,"passRate":0.85,"scoredRuns":20},"finance":{"outcome":0.8763,"passRate":0.75,"scoredRuns":20}}
{"premature-finalization":0.1616,"required-call-omitted":0.1616,"required-call-misordered":0.0202,"retry-abandoned":0,"required-write-missing":0.1515,"wrong-human-handoff":0.0404}
{"unauthorizedActions":0,"collateralChanges":27,"missedEscalations":4,"unnecessaryEscalations":0}
corpbench-work
1.3
corpbench-work-1-3
CorpBench Work v1.3
2026-09-13T21:01:59.321Z
2026-09-13T00:00:00
2026-09-13T00:00:00
c7a98c24e37dfe7e0d4f6cece54efb007293ce0c
https://github.com/RevensiAI/CorpBench/blob/work-v1.3/results/work/1.3/report.json
https://revensi.com/corpbench
100
5
18
1
custom-research
work
11
inception/mercury-2.5-preview
mercury-2-5-preview
Mercury 2.5 Preview
Inception
openrouter
false
75.7
0.9
0.4448
0.6
0.502003
0.690599
0.50975
0.69
0.6
0.002955
5.3275
41,570
4,817
1,513
852
100
100
60
60
0
0
0.001773
5.5
{"marketing":{"outcome":0.8512,"passRate":0.5,"scoredRuns":20},"sales":{"outcome":0.9504,"passRate":0.65,"scoredRuns":20},"revops":{"outcome":0.919,"passRate":0.55,"scoredRuns":20},"operations":{"outcome":0.8883,"passRate":0.5,"scoredRuns":20},"finance":{"outcome":0.8909,"passRate":0.8,"scoredRuns":20}}
{"premature-finalization":0.3061,"required-call-omitted":0.2959,"required-call-misordered":0.1735,"retry-abandoned":0.3636,"required-write-missing":0.102,"wrong-human-handoff":0.0204}
{"unauthorizedActions":0,"collateralChanges":28,"missedEscalations":2,"unnecessaryEscalations":0}
corpbench-work
1.3
corpbench-work-1-3
CorpBench Work v1.3
2026-09-13T21:01:59.321Z
2026-09-13T00:00:00
2026-09-13T00:00:00
c7a98c24e37dfe7e0d4f6cece54efb007293ce0c
https://github.com/RevensiAI/CorpBench/blob/work-v1.3/results/work/1.3/report.json
https://revensi.com/corpbench
100
5
18
1
custom-research
work
12
anthropic/claude-haiku-4.5
claude-haiku-4-5
Claude Haiku 4.5
Anthropic
openrouter
false
75.6
0.9153
0.1983
0.71
0.614611
0.789852
0.62
0.79
0.71
0.082338
19.092
50,192
0
1,590.5
862
100
100
71
71
0
0
0.05846
19.1
{"marketing":{"outcome":0.9827,"passRate":0.85,"scoredRuns":20},"sales":{"outcome":0.8734,"passRate":0.6,"scoredRuns":20},"revops":{"outcome":0.9741,"passRate":0.8,"scoredRuns":20},"operations":{"outcome":0.8713,"passRate":0.65,"scoredRuns":20},"finance":{"outcome":0.875,"passRate":0.65,"scoredRuns":20}}
{"premature-finalization":0.15,"required-call-omitted":0.12,"required-call-misordered":0.15,"retry-abandoned":0.1364,"required-write-missing":0.12,"wrong-human-handoff":0.01}
{"unauthorizedActions":0,"collateralChanges":58,"missedEscalations":1,"unnecessaryEscalations":0}
corpbench-work
1.3
corpbench-work-1-3
CorpBench Work v1.3
2026-09-13T21:01:59.321Z
2026-09-13T00:00:00
2026-09-13T00:00:00
c7a98c24e37dfe7e0d4f6cece54efb007293ce0c
https://github.com/RevensiAI/CorpBench/blob/work-v1.3/results/work/1.3/report.json
https://revensi.com/corpbench
100
5
18
1
custom-research
work
13
mistralai/mistral-small-2603
mistral-small-4
Mistral Small 4
Mistral AI
openrouter
false
74.8
0.8084
0.5498
0.72
0.62512
0.798603
0.62
0.81
0.72
0.001788
4.429
33,479
29,936
431.5
813
100
100
72
72
0
0
0.001287
4.5
{"marketing":{"outcome":0.74,"passRate":0.65,"scoredRuns":20},"sales":{"outcome":0.782,"passRate":0.6,"scoredRuns":20},"revops":{"outcome":0.889,"passRate":0.85,"scoredRuns":20},"operations":{"outcome":0.8361,"passRate":0.75,"scoredRuns":20},"finance":{"outcome":0.795,"passRate":0.75,"scoredRuns":20}}
{"premature-finalization":0.0964,"required-call-omitted":0.0964,"required-call-misordered":0.0482,"retry-abandoned":0,"required-write-missing":0.0241,"wrong-human-handoff":0}
{"unauthorizedActions":0,"collateralChanges":9,"missedEscalations":0,"unnecessaryEscalations":0}
corpbench-work
1.3
corpbench-work-1-3
CorpBench Work v1.3
2026-09-13T21:01:59.321Z
2026-09-13T00:00:00
2026-09-13T00:00:00
c7a98c24e37dfe7e0d4f6cece54efb007293ce0c
https://github.com/RevensiAI/CorpBench/blob/work-v1.3/results/work/1.3/report.json
https://revensi.com/corpbench
100
5
18
1
custom-research
work
14
thinkingmachines/inkling-small
inkling-small
Inkling Small
Thinking Machines Lab
openrouter
false
66.2
0.8225
0.2872
0.5
0.403832
0.596168
0.4
0.59025
0.5
0.016239
10.8985
53,885.5
48,064
523.5
857
100
100
50
50
0
0
0.008119
10.8
{"marketing":{"outcome":0.8729,"passRate":0.5,"scoredRuns":20},"sales":{"outcome":0.7328,"passRate":0.45,"scoredRuns":20},"revops":{"outcome":0.8789,"passRate":0.55,"scoredRuns":20},"operations":{"outcome":0.9205,"passRate":0.55,"scoredRuns":20},"finance":{"outcome":0.7073,"passRate":0.45,"scoredRuns":20}}
{"premature-finalization":0.4845,"required-call-omitted":0.134,"required-call-misordered":0.0619,"retry-abandoned":0.3636,"required-write-missing":0.4639,"wrong-human-handoff":0}
{"unauthorizedActions":0,"collateralChanges":85,"missedEscalations":0,"unnecessaryEscalations":0}
corpbench-work
1.3
corpbench-work-1-3
CorpBench Work v1.3
2026-09-13T21:01:59.321Z
2026-09-13T00:00:00
2026-09-13T00:00:00
c7a98c24e37dfe7e0d4f6cece54efb007293ce0c
https://github.com/RevensiAI/CorpBench/blob/work-v1.3/results/work/1.3/report.json
https://revensi.com/corpbench
100
5
18
1
custom-research
work
15
qwen/qwen3.8-flash
qwen3-8-flash
Qwen3.8 Flash
Qwen
openrouter
false
62.9
0.701
0.2861
0.66
0.562777
0.745385
0.56
0.75
0.66
0.005672
43.556
58,490.5
50,560
2,730.5
1,666
100
100
66
66
0
0
0.003743
60.4
{"marketing":{"outcome":0.7,"passRate":0.7,"scoredRuns":20},"sales":{"outcome":0.8327,"passRate":0.75,"scoredRuns":20},"revops":{"outcome":0.75,"passRate":0.75,"scoredRuns":20},"operations":{"outcome":0.5724,"passRate":0.45,"scoredRuns":20},"finance":{"outcome":0.65,"passRate":0.65,"scoredRuns":20}}
{"premature-finalization":0.0729,"required-call-omitted":0.0625,"required-call-misordered":0,"retry-abandoned":0.0476,"required-write-missing":0.0625,"wrong-human-handoff":0}
{"unauthorizedActions":0,"collateralChanges":14,"missedEscalations":0,"unnecessaryEscalations":0}
corpbench-work
1.3
corpbench-work-1-3
CorpBench Work v1.3
2026-09-13T21:01:59.321Z
2026-09-13T00:00:00
2026-09-13T00:00:00
c7a98c24e37dfe7e0d4f6cece54efb007293ce0c
https://github.com/RevensiAI/CorpBench/blob/work-v1.3/results/work/1.3/report.json
https://revensi.com/corpbench
100
5
18
1
custom-research
work
16
ollama/qwen3.5:9b
ollama-qwen3-5-9b
Qwen 3.5 9B (Ollama)
Qwen
ollama
true
62.4
0.8385
0.1742
0.38
0.290976
0.477902
0.28
0.48
0.38
0
44.1385
61,901.5
0
1,682
889
100
100
38
38
0
0
0
45.3
{"marketing":{"outcome":0.8748,"passRate":0.4,"scoredRuns":20},"sales":{"outcome":0.8822,"passRate":0.45,"scoredRuns":20},"revops":{"outcome":0.8648,"passRate":0.35,"scoredRuns":20},"operations":{"outcome":0.8627,"passRate":0.35,"scoredRuns":20},"finance":{"outcome":0.708,"passRate":0.35,"scoredRuns":20}}
{"premature-finalization":0.2959,"required-call-omitted":0.2959,"required-call-misordered":0.4898,"retry-abandoned":0.0455,"required-write-missing":0.0918,"wrong-human-handoff":0.051}
{"unauthorizedActions":0,"collateralChanges":36,"missedEscalations":5,"unnecessaryEscalations":0}
corpbench-work
1.3
corpbench-work-1-3
CorpBench Work v1.3
2026-09-13T21:01:59.321Z
2026-09-13T00:00:00
2026-09-13T00:00:00
c7a98c24e37dfe7e0d4f6cece54efb007293ce0c
https://github.com/RevensiAI/CorpBench/blob/work-v1.3/results/work/1.3/report.json
https://revensi.com/corpbench
100
5
18
1
custom-research
work
17
ollama/ministral-3:8b
ollama-ministral-3-8b
Ministral 3 8B (Ollama)
Mistral
ollama
true
46.7
0.6148
0.1881
0.28
0.201397
0.37488
0.19
0.37
0.28
0
14.293
28,232
0
522
792
100
100
28
28
0
0
0
14.8
{"marketing":{"outcome":0.7043,"passRate":0.4,"scoredRuns":20},"sales":{"outcome":0.6287,"passRate":0.3,"scoredRuns":20},"revops":{"outcome":0.6221,"passRate":0.15,"scoredRuns":20},"operations":{"outcome":0.5969,"passRate":0.4,"scoredRuns":20},"finance":{"outcome":0.522,"passRate":0.15,"scoredRuns":20}}
{"premature-finalization":0.5476,"required-call-omitted":0.5476,"required-call-misordered":0.3095,"retry-abandoned":0.2778,"required-write-missing":0.4643,"wrong-human-handoff":0.1071}
{"unauthorizedActions":0,"collateralChanges":50,"missedEscalations":6,"unnecessaryEscalations":3}
corpbench-work
1.3
corpbench-work-1-3
CorpBench Work v1.3
2026-09-13T21:01:59.321Z
2026-09-13T00:00:00
2026-09-13T00:00:00
c7a98c24e37dfe7e0d4f6cece54efb007293ce0c
https://github.com/RevensiAI/CorpBench/blob/work-v1.3/results/work/1.3/report.json
https://revensi.com/corpbench
100
5
18
1
custom-research
work
18
ollama/granite4.2:8b
ollama-granite4-2-8b
Granite 4.2 8B (Ollama)
IBM
ollama
true
31.1
0.4066
0.1165
0.2
0.133367
0.288829
0.12
0.28
0.2
0
17.919
53,398.5
0
603
794
100
100
20
20
0
0
0
17.8
{"marketing":{"outcome":0.466,"passRate":0.3,"scoredRuns":20},"sales":{"outcome":0.3788,"passRate":0.2,"scoredRuns":20},"revops":{"outcome":0.3683,"passRate":0.05,"scoredRuns":20},"operations":{"outcome":0.5346,"passRate":0.35,"scoredRuns":20},"finance":{"outcome":0.2853,"passRate":0.1,"scoredRuns":20}}
{"premature-finalization":0.2045,"required-call-omitted":0.2045,"required-call-misordered":0.0455,"retry-abandoned":1,"required-write-missing":0.1591,"wrong-human-handoff":0.3409}
{"unauthorizedActions":0,"collateralChanges":14,"missedEscalations":15,"unnecessaryEscalations":0}

CorpBench Work

CorpBench Work is a public library of 100 deterministic business workflows for evaluating AI agents across Finance, Marketing, Operations, RevOps, and Sales. This dataset is the Hugging Face release of CorpBench Work 1.3.

Explore the live CorpBench benchmark for the interactive leaderboard, findings, cost and speed comparisons, department results, failure modes, reliability, and methodology.

Dataset configurations

  • workflows contains 100 rows, one per workflow. It includes objectives, simulated systems, execution limits, agent-visible tool contracts, deterministic grader metadata, capabilities, version history, retirement policy, and withheld fixture identifiers.
  • results contains 18 rows, one per evaluated model in the published Work 1.3 cohort. It includes aggregate scores, confidence intervals, efficiency and reliability measures, department scores, failure rates, and safety counts.

Complex structures such as tool contracts and per-department scores are serialized as JSON strings in columns ending in _json. This keeps each row compatible with the Dataset Viewer without discarding information.

The exact public manifest and aggregate report are also preserved under release/, and their JSON Schemas are included under schemas/. The workflow rows were generated from GitHub commit c7a98c24e37dfe7e0d4f6cece54efb007293ce0c and carry each definition's manifest SHA-256 digest.

Usage 

from datasets import load_dataset

workflows = load_dataset(
    "revensi/CorpBench",
    "workflows",
    split="test",
    revision="work-v1.3",
)

results = load_dataset(
    "revensi/CorpBench",
    "results",
    split="test",
    revision="work-v1.3",
)

Coverage

Department Workflows
Finance 20
Marketing 20
Operations 20
RevOps 20
Sales 20

Evaluation boundary

This is a public workflow library and aggregate-data release, not a standalone benchmark runner or an uncontaminated private holdout. Seeded workspaces, expected states, reference calls, evaluator implementation, reproduction snapshots, raw traces, run IDs, provider configuration, and private research notes are intentionally withheld.

Public workflow definitions may appear in model training data after release. Official CorpBench evaluations use separately controlled fixtures and evaluator infrastructure.

Versioning

The immutable revision for this release is work-v1.3. Future Work releases use work-v<version>. Machine identifiers remain lowercase (corpbench-work) while display branding remains CorpBench.

License and citation

The workflow definitions, metadata, schemas, and aggregate data are released under Apache-2.0. Citation metadata is included in CITATION.cff.

RevensiAI. CorpBench Work 1.3. 2026. https://github.com/RevensiAI/CorpBench
Downloads last month
77
Number of rows:
118
Total file size:
2.59 MB