Recent benchmark leadership in agentic tasks such as Terminal-Bench, OSWorld-Verified, and BrowseComp remains fragmented, with Shanghai AI Lab’s Atria Dawn Preview, OpenAI’s GPT-5.6/GPT-6 series, Moonshot’s Kimi K3, Anthropic’s Claude Opus 5, and DeepSeek models trading top spots across coding, browsing, and multi-step workflows. This fluidity, combined with strong revenue traction at Cognition (Devin) and Sierra alongside rapid open-weight progress from Z.ai, Alibaba Qwen, and Moonshot, sustains broad trader uncertainty over which lab will rank third by late November. Differentiators include demonstrated long-horizon reliability, enterprise deployment scale, and post-training optimizations for tool use rather than raw model size. Upcoming model releases and benchmark updates through October could quickly shift the crowded field.
基於Polymarket數據的AI實驗性摘要。這不是交易建議,也不影響該市場的結算方式。 · 更新於Moonshot 28%
Z.ai 23%
Anthropic 22%
OpenAI 22%

Moonshot
28%

Z.ai
23%

Anthropic
22%

OpenAI
22%

22%

SpaceXAI
18%

美團
14%

DeepSeek
14%

字節跳動
14%

亞馬遜
12%

MiniMax
12%

騰訊
12%

阿里巴巴
12%

Nvidia
12%

Mistral
12%

百度
11%

小米
11%

微軟
11%

Meta
9%
Moonshot 28%
Z.ai 23%
Anthropic 22%
OpenAI 22%

Moonshot
28%

Z.ai
23%

Anthropic
22%

OpenAI
22%

22%

SpaceXAI
18%

美團
14%

DeepSeek
14%

字節跳動
14%

亞馬遜
12%

MiniMax
12%

騰訊
12%

阿里巴巴
12%

Nvidia
12%

Mistral
12%

百度
11%

小米
11%

微軟
11%

Meta
9%
Results from the "Rank" column under the "Agent Arena" Leaderboard tab at https://arena.ai/leaderboard/agent filtered for "Labs" will be used to resolve this market.
Models marked “AutoEval” at the applicable check time will not be considered, regardless of whether they display a rank or score.
AI companies will be ordered primarily by their Lab Rank at the market’s check time. If the results based on the lab ranking are ambiguous or unavailable, the relevant AI companies will be ordered according to their highest-ranking AI model in the leaderboard’s “Models” view. If two or more models are tied on rank, they will be ordered by which model is listed higher on the leaderboard. If a tie still remains, alphabetical order of AI lab/company names as listed in this market group will be used as a final tiebreaker (e.g., “Google” would be ranked ahead of “SpaceXAI”). This market will resolve based on the company that occupies third place under this ranking.
The resolution source for this market is the arena.ai Agent Arena Leaderboard. If this resolution source is unavailable at check time, this market will remain open until the leaderboard comes back online and will resolve based on the first check after it becomes available. If it becomes permanently unavailable, this market will resolve to "Other".
市場開放時間: Sep 17, 2026, 8:04 PM ET
Results from the "Rank" column under the "Agent Arena" Leaderboard tab at https://arena.ai/leaderboard/agent filtered for "Labs" will be used to resolve this market.
Models marked “AutoEval” at the applicable check time will not be considered, regardless of whether they display a rank or score.
AI companies will be ordered primarily by their Lab Rank at the market’s check time. If the results based on the lab ranking are ambiguous or unavailable, the relevant AI companies will be ordered according to their highest-ranking AI model in the leaderboard’s “Models” view. If two or more models are tied on rank, they will be ordered by which model is listed higher on the leaderboard. If a tie still remains, alphabetical order of AI lab/company names as listed in this market group will be used as a final tiebreaker (e.g., “Google” would be ranked ahead of “SpaceXAI”). This market will resolve based on the company that occupies third place under this ranking.
The resolution source for this market is the arena.ai Agent Arena Leaderboard. If this resolution source is unavailable at check time, this market will remain open until the leaderboard comes back online and will resolve based on the first check after it becomes available. If it becomes permanently unavailable, this market will resolve to "Other".
Recent benchmark leadership in agentic tasks such as Terminal-Bench, OSWorld-Verified, and BrowseComp remains fragmented, with Shanghai AI Lab’s Atria Dawn Preview, OpenAI’s GPT-5.6/GPT-6 series, Moonshot’s Kimi K3, Anthropic’s Claude Opus 5, and DeepSeek models trading top spots across coding, browsing, and multi-step workflows. This fluidity, combined with strong revenue traction at Cognition (Devin) and Sierra alongside rapid open-weight progress from Z.ai, Alibaba Qwen, and Moonshot, sustains broad trader uncertainty over which lab will rank third by late November. Differentiators include demonstrated long-horizon reliability, enterprise deployment scale, and post-training optimizations for tool use rather than raw model size. Upcoming model releases and benchmark updates through October could quickly shift the crowded field.
基於Polymarket數據的AI實驗性摘要。這不是交易建議,也不影響該市場的結算方式。 · 更新於
警惕外部連結哦。
警惕外部連結哦。
Frequently Asked Questions