TUFA Labs Tops ARC-AGI-2 with 88.06% · 1010-726

2026-10-10 00:00

Author: RadarAI Editorial Editor: RadarAI Editorial Last updated: 2026-10-10 Review status: Editorial review pending Brief 速报官方 AI动态开源

Editorial standards and source policy: Editorial standards, Team. Content links to primary sources; see Methodology.

## 🔍 Core Insights **ARC-AGI-2** competition has seen a historic breakthrough, with **TUFA Labs** topping the leaderboard at **88.06%**. More than 85% of teams will share a $150,000 prize pool, showing that reasoning models are rapidly approaching human-level abstract generalization [1][3]. Meanwhile, **ElevenLabs**' partnership with **Banner Health** marks AI voice agents officially entering core healthcare workflows, while **Sabi**'s $50 million funding round signals that brain-computer interfaces are moving from the lab toward consumer-grade wearables [0][12]. ## 🚀 Key Updates - **ARC-AGI-2 competition sets new record** [1][3]: TUFA Labs tops the leaderboard at 88.06%, over 85% of teams share a $150,000 prize pool, reasoning ability approaches human level. - **ElevenLabs voice agents land in healthcare** [0]: Partnership with Banner Health, AI directly connects to electronic medical records, achieving full automation of scheduling and rescheduling. - **Sabi raises $50 million seed round** [12]: Developing a brain-computer interface baseball cap that converts brain signals into AI prompts, led by Vinod Khosla, valued at $600 million. - **OpenAI launches MXC sandbox for Windows Codex** [10]: Based on Microsoft Execution Containers, providing faster setup and stronger network isolation. - **Xiaomi MiMo model iteration** [11]: MiMo-V2.5 series discontinued, recommended migration to MiMo V2.6 series with stronger reasoning and better cost-effectiveness. - **Unsloth releases free decision model training Notebook** [13]: Can locally train Qwen3.5-4B into a decision model, requiring only 8GB VRAM. - **Arena leaderboard update** [14]: Nano Banana 2.1 enters top six in three Image Arena modes, ranks 4th in Multi-Image Edit. - **Deno acquired by Cloudflare** [24]: Runtime project will shut down after one year of operation, reflecting the shifting competitive landscape of open-source frameworks in the AI era. ## 🔗 Sources [0] ElevenLabs partners with Banner Health — https://aihot.news/items/c9t5r4t3ld2ca5qov7zl0ac6c [1] François Chollet discusses ARC-AGI-2 score reaching 88.06% on Kaggle — https://aihot.news/items/yissvbkrdl19v6hmgeekyw7v9 [3] ARC Prize 2026: TUFA Labs tops ARC-AGI-2 high score leaderboard at 88.06% — https://aihot.news/items/o8tu2a2i9r2roqh7un9l5grq2 [10] OpenAI launches new MXC-based sandbox mode for Codex on Windows — https://aihot.news/items/cz35f64wgq0vaxdpv2zaaq8id [11] Xiaomi MiMo-V2.5-Pro and MiMo-V2.5 models discontinued, can migrate to MiMo V2.6 series — https://aihot.news/items/jvn7sezxoc1h6zpa1wtox69id [12] Sabi raises $50 million seed round, developing brain-computer interface baseball cap that converts brain signals into AI prompts — https://aihot.news/items/tp1nvmmi18f1aakergjbatu5

ARC-AGI-2 competition has seen a historic breakthrough, with TUFA Labs topping the leaderboard at 88.06%. More than 85% of teams will share a $150,000 prize pool, showing that reasoning models are rapidly approaching human-level abstract generalization [1][3]. Meanwhile, ElevenLabs' partnership with Banner Health marks AI voice agents officially entering core healthcare workflows, while Sabi's $50 million funding round signals that brain-computer interfaces are moving from the lab toward consumer-grade wearables [0][12].

🚀 Key Updates

ARC-AGI-2 competition sets new record [1][3]: TUFA Labs tops the leaderboard at 88.06%, over 85% of teams share a $150,000 prize pool, reasoning ability approaches human level.
ElevenLabs voice agents land in healthcare [0]: Partnership with Banner Health, AI directly connects to electronic medical records, achieving full automation of scheduling and rescheduling.
Sabi raises $50 million seed round [12]: Developing a brain-computer interface baseball cap that converts brain signals into AI prompts, led by Vinod Khosla, valued at $600 million.
OpenAI launches MXC sandbox for Windows Codex [10]: Based on Microsoft Execution Containers, providing faster setup and stronger network isolation.
Xiaomi MiMo model iteration [11]: MiMo-V2.5 series discontinued, recommended migration to MiMo V2.6 series with stronger reasoning and better cost-effectiveness.
Unsloth releases free decision model training Notebook [13]: Can locally train Qwen3.5-4B into a decision model, requiring only 8GB VRAM.
Arena leaderboard update [14]: Nano Banana 2.1 enters top six in three Image Arena modes, ranks 4th in Multi-Image Edit.
Deno acquired by Cloudflare [24]: Runtime project will shut down after one year of operation, reflecting the shifting competitive landscape of open-source frameworks in the AI era.

🔗 Sources

[0] ElevenLabs partners with Banner Health — https://aihot.news/items/c9t5r4t3ld2ca5qov7zl0ac6c [1] François Chollet discusses ARC-AGI-2 score reaching 88.06% on Kaggle — https://aihot.news/items/yissvbkrdl19v6hmgeekyw7v9 [3] ARC Prize 2026: TUFA Labs tops ARC-AGI-2 high score leaderboard at 88.06% — https://aihot.news/items/o8tu2a2i9r2roqh7un9l5grq2 [10] OpenAI launches new MXC-based sandbox mode for Codex on Windows — https://aihot.news/items/cz35f64wgq0vaxdpv2zaaq8id [11] Xiaomi MiMo-V2.5-Pro and MiMo-V2.5 models discontinued, can migrate to MiMo V2.6 series — https://aihot.news/items/jvn7sezxoc1h6zpa1wtox69id [12] Sabi raises $50 million seed round, developing brain-computer interface baseball cap that converts brain signals into AI prompts — https://aihot.news/items/tp1nvmmi18f1aakergjbatu5

← Back to Updates