LocalAgent
Benchmark

Apache-2.0
  1. 01
    smoke

    1 task
  2. 02
    standard

    4 task
  3. 03
    full

    5 task

01 —

01

02

03

04

80%
10%
5%
5%

02 —

  1. 01

  2. 02

  3. 03

  4. 04

03 —

  1. smoke

  2. 4standard

  3. 2full

04 —

05 —

06 —

localagent-benchmark
$ python3 benchmark.py run --profile smoke \
    --models MODEL_A MODEL_B --thinking off
$ python3 benchmark.py run --profile standard \
    --models MODEL_A --seed 42 --sandbox required
$ python3 benchmark.py run --profile full \
    --models MODEL_A --repetitions 3
$ python3 benchmark.py run --cases CASE_ID \
    --models MODEL_A --timeout 1200
$ python3 benchmark.py run --profile showcase \
    --models FINALIST_A FINALIST_B --thinking off

07 —

08 —

INSTRUCTIONS.md QUICK-START_Guided.md QUICK-START_Dashboard.md QUICK-START_Showcase.md QUICK-START_Case-Author.md QUICK-START_Windows.md QUICK-START_Linux.md