Baseline01
Can the agent complete the task?
A broad collection of bioinformatics tasks with explicit deliverables and verification rules.
BioinfoAgentBench
A benchmark for bioinformatics agents, covering task completion, PARE capabilities, and scientific research.
01
Baseline
Task completion
810tasks
02
Fingerprint
PARE capabilities
100tasks
03
Frontier
Scientific problem solving
5cases
7Task sources
8PARE subdimensions
6Evaluated agents
Benchmark tiers
Fingerprint02
What are its strengths and limitations?
A capability profile across Planning, Acting, Rectifying, and Evolving, grounded in calibrated task loads.
Frontier03
How far can it take a scientific question?
Research-oriented challenges connecting analytical workflows, reference findings, and numerical evidence.
PARE capabilities
Planning
Decompose a scientific goal and choose appropriate methods.
Acting
Generate analytical results and follow the stated requirements.
Rectifying
Resolve ambiguous instructions and check the supplied data.
Evolving
Use experience and adapt strategies to a new task.
Results
View resultsBaseline
| Agent | Correctness | Resource use | |||||
|---|---|---|---|---|---|---|---|
| B1 · Success rate | B2 · Subject macro-average | B3 · Token usage (k/task) | B4 · Execution time (min/task) | B5 · Interaction turns (/task) | B6 · Cost (USD/task) | B7 · Cost (USD/pass) | |
| AutoBA | 0.34 | 0.39 | 72.38 | 34.41 | 27.10 | 0.247 | 0.726 |
| BioMaster | 0.05 | 0.11 | 68.92 | 14.81 | 7.77 | 0.267 | 5.340 |
| Biomni | 0.78 | 0.76 | 243.65 | 3.70 | 5.62 | 0.116 | 0.148 |
| BixBench-Agent | 0.76 | 0.76 | 40.76 | 4.56 | 6.65 | 0.128 | 0.168 |
| STELLA | 0.71 | 0.74 | 367.81 | 6.26 | 20.23 | 0.247 | 0.347 |
| BioMedAgent | 0.73 | 0.72 | 167.57 | 10.37 | 13.21 | 0.412 | 0.565 |
PARE fingerprint
| Agent | B6 · Cost (USD/task) | Planning | Acting | Rectifying | Evolving | ||||
|---|---|---|---|---|---|---|---|---|---|
| P1 | P2 | A1 | A2 | R1 | R2 | E1 | E2 | ||
| AutoBA | 0.569 | 0.133 | 0.215 | 0.352 | 0.400 | 0.440 | 0.125 | 0.121 | 0.243 |
| BioMaster | 0.272 | 0.000 | 0.000 | 0.028 | 0.024 | 0.200 | 0.000 | 0.000 | 0.000 |
| Biomni | 0.105 | 0.644 | 0.709 | 0.789 | 0.800 | 0.680 | 0.750 | 0.707 | 0.757 |
| BixBench-Agent | 0.122 | 0.844 | 0.886 | 0.887 | 0.904 | 0.600 | 0.750 | 0.759 | 0.919 |
| STELLA | 0.302 | 0.756 | 0.772 | 0.803 | 0.824 | 0.760 | 0.188 | 0.621 | 0.649 |
| BioMedAgent | 1.167 | 0.822 | 0.899 | 0.831 | 0.856 | 0.800 | 0.750 | 0.655 | 0.838 |
Frontier
| Agent | F1 · Approaching | F2 · Reproducing | F3 · Exceeding |
|---|---|---|---|
| AutoBA | 0.375 | 0.000 | 0.000 |
| BioMaster | 0.400 | 0.015 | 0.000 |
| Biomni | 0.800 | 0.090 | 0.067 |
| BixBench-Agent | 0.650 | 0.107 | 0.067 |
| STELLA | 0.725 | 0.199 | 0.067 |
| BioMedAgent | 0.850 | 0.199 | 0.133 |