Enterprise Interface Task Completion on WorkArena L1
79.7Task Success RateGemini 3 Pro
Evaluation Results
| Method | Links | |
|---|---|---|
| Gemini 3 ProCategory=Proprietary, Teacher for A3-SYNTH=true2026.04 | 79.7 | |
| GPT-5Harness=GenericAgent, Params=-2026.04 | 79.1 | |
| Claude 4 SonnetHarness=GenericAgent, Params=-2026.04 | 63.3 | |
| Gemini 3.1 Flash L.Category=Proprietary2026.04 | 58.5 | |
| Qwen3.5-27BCategory=Open-weight (base)2026.04 | 57 | |
| A3-Qwen3.5-9BCategory=A3 fine-tuned, Teacher=Gemini 3 Pro2026.04 | 51.5 | |
| A3-Qwen3.5-9BHarness=GenericAgent, Params=9B2026.04 | 51.5 | |
| GPT-oss-120BHarness=GenericAgent, Params=120B2026.04 | 50.9 | |
| A3-Qwen3.5-4BCategory=A3 fine-tuned, Teacher=Gemini 3 Pro2026.04 | 44.8 | |
| Qwen3.5-4BCategory=Open-weight (base)2026.04 | 33.6 | |
| Qwen3.5-9BCategory=Open-weight (base)2026.04 | 33.3 | |
| Qwen3.5-9B (base)Harness=GenericAgent, Params=9B2026.04 | 33.3 | |
| A3-Qwen3.5-2BCategory=A3 fine-tuned, Teacher=Gemini 3 Pro2026.04 | 6.7 | |
| Qwen3.5-2BCategory=Open-weight (base)2026.04 | 4.2 |