Omni-modal agentic tasks on OmniGAIA
75Pass@1 (Overall)GPT-5.4 xHigh
Evaluation Results
| Method | Links | |||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| GPT-5.4 xHighHarness=Codex, Reasoning effort=xHigh2026.05 | 75 | — | — | — | — | — | — | — | — | — | 82 | 75 | 64.1 | |
| Orchestra-o1-GPT-5Model Accessibility=Proprietary, Backbone=GPT-5, Agent Framework=Orchestra-o12026.06 | 72.8 | 72.5 | 69.4 | 75.8 | 64 | 83.8 | 63.9 | 69.7 | 73.1 | 83.3 | — | — | — | |
| Claude Opus 4.6Harness=CC2026.05 | 68.6 | — | — | — | — | — | — | — | — | — | 74.6 | 69.4 | 57.7 | |
| OmniAgentBackbone=GPT-5.4 High2026.05 | 66.4 | — | — | — | — | — | — | — | — | — | 77.9 | 63.1 | 55.1 | |
| Gemini 3.1 Pro2026.05 | 66.1 | — | — | — | — | — | — | — | — | — | 77.8 | 63.8 | 52.6 | |
| GPT-5.4 High§Harness=Codex, Direct image read=disabled2026.05 | 64.4 | — | — | — | — | — | — | — | — | — | 70.5 | 63.8 | 56.4 | |
| Claude Sonnet 4.6Harness=CC2026.05 | 63.3 | — | — | — | — | — | — | — | — | — | 73.8 | 62.5 | 48.8 | |
| Gemini-3-Pro# Params=-2026.02 | 62.5 | 65.2 | 59.2 | 62.1 | 72 | 78.4 | 52.8 | 48.5 | 42.3 | 88.9 | 78.7 | 61.9 | 38.5 | |
| Gemini-3-ProModel Accessibility=Proprietary, Agent Framework=ReAct2026.06 | 62.5 | 65.2 | 59.2 | 62.1 | 72 | 78.4 | 52.8 | 48.5 | 42.3 | 88.9 | — | — | — | |
| GPT-5.4 HighHarness=Codex, Reasoning effort=High2026.05 | 61.4 | — | — | — | — | — | — | — | — | — | 70.4 | 60 | 50 | |
| GPT-5.4 MediumHarness=Codex, Reasoning effort=Medium2026.05 | 56.4 | — | — | — | — | — | — | — | — | — | 66.4 | 55 | 43.6 | |
| Kimi K-2.5Harness=CC2026.05 | 55.6 | — | — | — | — | — | — | — | — | — | 72.1 | 50.6 | 39.7 | |
| GPT-5.4 LowHarness=Codex, Reasoning effort=Low2026.05 | 55 | — | — | — | — | — | — | — | — | — | 63.1 | 53.8 | 44.9 | |
| Codex-GPT-5.4 LowSetting=Strong Baseline2026.05 | 55 | — | — | — | — | — | — | — | — | — | 63.1 | 53.8 | 44.9 | |
| Gemini-3-Flash# Params=-2026.02 | 51.7 | 50.7 | 57.1 | 44.8 | 48 | 59.5 | 55.6 | 54.6 | 38.5 | 61.1 | 67.2 | 46.9 | 37.2 | |
| Gemini 3 Flash2026.05 | 51.7 | — | — | — | — | — | — | — | — | — | 67.2 | 46.9 | 37.2 | |
| Gemini-3-FlashModel Accessibility=Proprietary, Agent Framework=ReAct2026.06 | 51.7 | 50.7 | 57.1 | 44.8 | 48 | 59.5 | 55.6 | 54.6 | 38.5 | 61.1 | — | — | — | |
| Code-X (SFT)Model=Qwen3.6-27B, Thinking budget=4K2026.05 | 43.3 | — | — | — | — | — | — | — | — | — | 50 | 45 | 28.2 | |
| Agent-OmniBackbone=GPT-5.4 High2026.05 | 41.1 | — | — | — | — | — | — | — | — | — | 54.1 | 41.9 | 19.2 | |
| AOrchestra-GPT-5Model Accessibility=Proprietary, Backbone=GPT-5, Agent Framework=AOrchestra2026.06 | 40 | 34.8 | 40.8 | 56.1 | 32 | 51.4 | 25 | 42.4 | 30.8 | 22.2 | — | — | — | |
| Qwen 3.5 Omni Plus2026.05 | 38.7 | — | — | — | — | — | — | — | — | — | 43.5 | 39.4 | 29 | |
| Direct MLLMModel=Qwen3.6-27B, Thinking budget=4K2026.05 | 38.2 | — | — | — | — | — | — | — | — | — | 44.3 | 38.1 | 28.6 | |
| MiniMax-M2.7‡Harness=CC, Note=Uses own MCP tool for image reading2026.05 | 33.3 | — | — | — | — | — | — | — | — | — | 38.5 | 34.4 | 23.1 | |
| Gemini-2.5-Pro# Params=-2026.02 | 30.8 | 23.2 | 28.6 | 32.8 | 20 | 32.4 | 41.7 | 42.4 | 26.9 | 33.3 | 41.8 | 26.9 | 21.8 | |
| Gemini-2.5-ProModel Accessibility=Proprietary, Agent Framework=ReAct2026.06 | 30.8 | 23.2 | 28.6 | 32.8 | 20 | 32.4 | 41.7 | 42.4 | 26.9 | 33.3 | — | — | — | |
| Orchestra-o1-8BModel Accessibility=Open-Source, Backbone=Qwen3-8B, Agent Framework=Orchestra-o12026.06 | 30 | 21.7 | 32.7 | 37.9 | 12 | 29.7 | 16.7 | 45.5 | 38.5 | 38.9 | — | — | — | |
| Code-X (RL)Model=Qwen3.5-9B, Thinking budget=4K2026.05 | 28.1 | — | — | — | — | — | — | — | — | — | 32.8 | 26.3 | 24.4 | |
| Mimo V2 Omni2026.05 | 25.8 | — | — | — | — | — | — | — | — | — | 33.9 | 22.5 | 19.1 | |
| Baseline (direct image)Model=Qwen3.6-27B, Thinking budget=4K2026.05 | 24.4 | — | — | — | — | — | — | — | — | — | 32.8 | 18.8 | 23.1 | |
| Code-X (SFT)Model=Qwen3.5-9B, Thinking budget=4K2026.05 | 23.1 | — | — | — | — | — | — | — | — | — | 30 | 21 | 17 | |
| OmniAtlas-Qwen-3# Params=30B-A3B2026.02 | 20.8 | 10.1 | 30.6 | 29.9 | 32 | 18.9 | 16.7 | 12.1 | 11.5 | 27.8 | 31.1 | 18.8 | 9 | |
| OmniAtlas-Omni-30B-A3BSetting=Strong Baseline2026.05 | 20.8 | — | — | — | — | — | — | — | — | — | 31.1 | 18.8 | 9 | |
| OmniAtlas-Qwen3-30B-A3BModel Accessibility=Open-Source, Agent Framework=ReAct2026.06 | 20.8 | 10.1 | 30.6 | 29.9 | 32 | 18.9 | 16.7 | 12.1 | 11.5 | 27.8 | — | — | — | |
| Direct MLLMModel=Qwen3.5-9B, Thinking budget=4K2026.05 | 20 | — | — | — | — | — | — | — | — | — | 23 | 16.2 | 23.1 | |
| Baseline (direct image)Model=Qwen3.5-9B, Thinking budget=4K2026.05 | 18.3 | — | — | — | — | — | — | — | — | — | 23.8 | 14.4 | 17.9 | |
| Qwen-3-Omni# Params=30B-A3B2026.02 | 13.3 | 8.7 | 14.3 | 11.9 | 28 | 10.8 | 13.9 | 9.1 | 15.4 | 22.2 | 19.7 | 10.6 | 9 | |
| OmniAtlas-Qwen-2.5# Params=7B2026.02 | 13.3 | 8.7 | 18.4 | 16.4 | 4 | 16.2 | 22.2 | 3 | 7.7 | 22.2 | 22.1 | 11.3 | 3.9 | |
| Qwen 3 Omni2026.05 | 13.3 | — | — | — | — | — | — | — | — | — | 19.7 | 10.6 | 9 | |
| Qwen3-Omni-30B-A3BModel Accessibility=Open-Source, Agent Framework=ReAct2026.06 | 13.3 | 8.7 | 14.3 | 11.9 | 28 | 10.8 | 13.9 | 9.1 | 15.4 | 22.2 | — | — | — | |
| OmniAtlas-Qwen2.5-7BModel Accessibility=Open-Source, Agent Framework=ReAct2026.06 | 13.3 | 8.7 | 18.4 | 16.4 | 4 | 16.2 | 22.2 | 3 | 7.7 | 22.2 | — | — | — | |
| LongCat-Flash-Omni# Params=560B-A27B2026.02 | 11.1 | 8.7 | 10.2 | 16.4 | 12 | 10.8 | 8.3 | 6.1 | 11.5 | 16.7 | 16.4 | 9.4 | 6.4 | |
| LongCat-Flash-Omni-560B-A27BModel Accessibility=Open-Source, Agent Framework=ReAct2026.06 | 11.1 | 8.7 | 10.2 | 16.4 | 12 | 10.8 | 8.3 | 6.1 | 11.5 | 16.7 | — | — | — | |
| OmniAtlas-Qwen-2.5# Params=3B2026.02 | 10.3 | 4.4 | 12.2 | 16.7 | 4 | 16.2 | 11.1 | 3 | 11.5 | 11.1 | 13.9 | 10 | 5.1 | |
| OmniAtlas-Qwen2.5-3BModel Accessibility=Open-Source, Agent Framework=ReAct2026.06 | 10.3 | 4.4 | 12.2 | 16.7 | 4 | 16.2 | 11.1 | 3 | 11.5 | 11.1 | — | — | — | |
| Gemini-2.5-Flash-Lite# Params=-2026.02 | 8.6 | 5.8 | 8.2 | 14.9 | 4 | 10.8 | 8.3 | 6.1 | 3.9 | 11.1 | 9.8 | 8.1 | 7.7 | |
| Gemini-2.5-Flash-LiteModel Accessibility=Proprietary, Agent Framework=ReAct2026.06 | 8.6 | 5.8 | 8.2 | 14.9 | 4 | 10.8 | 8.3 | 6.1 | 3.9 | 11.1 | — | — | — | |
| Ming-Flash-Omni# Params=100B-A6B2026.02 | 8.3 | 5.8 | 8.2 | 10.4 | 12 | 8.1 | 5.6 | 6.1 | 11.5 | 11.1 | 12.3 | 7.5 | 3.8 | |
| Ming-Flash-Omni-100B-A6BModel Accessibility=Open-Source, Agent Framework=ReAct2026.06 | 8.3 | 5.8 | 8.2 | 10.4 | 12 | 8.1 | 5.6 | 6.1 | 11.5 | 11.1 | — | — | — | |
| Ming-Lite-Omni-1.5# Params=20B-A3B2026.02 | 3.9 | 2.9 | 6.1 | 1.5 | 4 | 5.4 | 2.8 | 6.1 | 7.7 | 5.6 | 4.9 | 3.8 | 2.6 | |
| Ming-Lite-Omni-1.5-20B-A3BModel Accessibility=Open-Source, Agent Framework=ReAct2026.06 | 3.9 | 2.9 | 6.1 | 1.5 | 4 | 5.4 | 2.8 | 6.1 | 7.7 | 5.6 | — | — | — | |
| Qwen-2.5-Omni# Params=7B2026.02 | 3.6 | 1.5 | 4.1 | 7.5 | 4 | 0 | 2.8 | 0 | 7.7 | 5.6 | 8.2 | 1.3 | 1.3 | |
| Qwen2.5-Omni-7BModel Accessibility=Open-Source, Agent Framework=ReAct2026.06 | 3.6 | 1.5 | 4.1 | 7.5 | 4 | 0 | 2.8 | 0 | 7.7 | 5.6 | — | — | — | |
| MiniCPM-O-2.6# Params=8B2026.02 | 3.1 | 2.9 | 2 | 1.5 | 0 | 2.7 | 8.3 | 3 | 3.8 | 5.6 | 3.3 | 2.5 | 3.8 | |
| MiniCPM-O-2.6-8BModel Accessibility=Open-Source, Agent Framework=ReAct2026.06 | 3.1 | 2.9 | 2 | 1.5 | 0 | 2.7 | 8.3 | 3 | 3.8 | 5.6 | — | — | — | |
| Baichuan-Omni-1.5# Params=8B2026.02 | 2.8 | 2.9 | 4.1 | 3 | 4 | 2.7 | 0 | 3 | 3.8 | 0 | 4.9 | 2.5 | 0 | |
| Baichuan-Omni-1.5-8BModel Accessibility=Open-Source, Agent Framework=ReAct2026.06 | 2.8 | 2.9 | 4.1 | 3 | 4 | 2.7 | 0 | 3 | 3.8 | 0 | — | — | — | |
| Qwen-2.5-Omni# Params=3B2026.02 | 1.4 | 0 | 2 | 4.5 | 0 | 0 | 0 | 0 | 3.9 | 0 | 1.6 | 1.9 | 0 | |
| Qwen2.5-Omni-3BModel Accessibility=Open-Source, Agent Framework=ReAct2026.06 | 1.4 | 0 | 2 | 4.5 | 0 | 0 | 0 | 0 | 3.9 | 0 | — | — | — |