AI News

Trending story

ARC Prize announces the results of three ARC-AGI evaluations of Inkling 2 Small

1 reports1 story updatesUpdated Nov 17 05:29

Understand this story

AI overview

ARC Prize announces the verified results of Thinking Machines' Inkling 2 Small on three ARC-AGI benchmarks. It scored 81.7% on ARC-AGI-2 at $0.27 per task, and 96.5% on ARC-AGI-1 at $0.08 per task; ARC Prize states that the model surpasses all open-weight models it has evaluated on these two benchmarks. On ARC-AGI-3, the model scored 2.6% under the provider adapter ($704 total) and 0.3% under standard ($858 total). Compared with DeepSeek V4.1 Flash 1017, Inkling 2 Small leads on ARC-AGI-1 (96.5% vs. 94.5%) and ARC-AGI-2 (81.7% vs. 78.1%), while trailing on ARC-AGI-3 (2.6% vs. 5.8%). Thinking Machines plans to release the full weights next week.

AI-generated from public reports

Latest development · Report

ARC Prize announces Inkling 2 Small's results on three ARC-AGI benchmarks

Developments

1 confirmed developments

  1. ARC Prize announces evaluation results of Inkling 2 Small on ARC-AGI

    Source:ARC Prize

Reports and discussion timeline

Browse reports and discussion by time. Their evidence levels remain distinct.

Nov 17

  1. ARC PrizeFirst-partySelected

    ARC Prize announces Inkling 2 Small's results on three ARC-AGI benchmarks

    ARC Prize has published the verified results of Thinking Machines' Inkling 2 Small on ARC-AGI: it scored 81.7% on ARC-AGI-2 at $0.27 per task, and 96.5% on ARC-AGI-1 at $0.08 per task, surpassing every open-weight model it has evaluated on both benchmarks. On ARC-AGI-3, the model scored 2.6% (provider adapter, $704 total) and 0.3% (standard, $858 total). Compared with DeepSeek V4.1 Flash 1017, it leads on ARC-AGI-1 (96.5% vs. 94.5%) and ARC-AGI-2 (81.7% vs. 78.1%), but trails on ARC-AGI-3 (2.6% vs. 5.8%). Thinking Machines plans to release the full weights next week.

Related stories

  1. Related storiesQwen Releases Qwen4-Max Preview Version
  2. Related storiesMoonshot AI releases Kimi K3.1 Pro model weights
  3. Related storiesMistral releases Large 4 weights on Hugging Face
  4. Related storiesCognition Launches Coding Benchmark FrontierCode 2.0