TuringPost

10 notable AI models of the week:

▪️ Aya Vision
▪️ INTELLECT-2
▪️ MiniMax-Speech
▪️ SWE-1
▪️ Seed1.5-VL
▪️ BLIP3-o
▪️ Skywork-VL
▪️ Behind Maya
▪️ MiMo
▪️ AM-Thinking-v1

🧵 

1. Aya Vision by @cohere

Scales multilingual multimodal generation using synthetic annotation and cross-modal merging, outperforming larger models in multilingual visual reasoning tasks.

https://arxiv.org/abs/2505.08751

Models
Aya-Vision-8B: https://huggingface.co/CohereLabs/aya-vision-8B
Aya-Vision-32B: https://huggingface.co/CohereLabs/aya-vision-32B

2. INTELLECT-2 by Prime Intellect Team

Trains a reasoning model via globally decentralized reinforcement learning, innovating with asynchronous infrastructure and outperforming the 32B reasoning baseline.

https://arxiv.org/abs/2505.07291 

3. MiniMax-Speech by @MiniMax__AI

Achieves zero-shot high-quality speech synthesis with a learnable speaker encoder and Flow-VAE, excelling in cloning, emotion control, and multilingual support.

https://arxiv.org/abs/2505.07916

4. SWE-1 by @windsurf_ai

Accelerate software engineering workflows by integrating tool-aware reasoning, incomplete state tracking, and timeline-based context for superior agentic coding experiences.

https://windsurf.com/blog/windsurf-wave-9-swe-1

5. Seed1.5-VL by @ByteDanceOSS

Advances agent-centric multimodal reasoning with a compact vision encoder and MoE-based LLM that outperforms competitors on 38 of 60 benchmarks and leads in GUI and gameplay tasks.

https://arxiv.org/abs/2505.07062 

6. BLIP3-o by @salesforce

Unifies image understanding and generation by using a diffusion transformer for CLIP feature generation and sequential training that preserves comprehension while enabling generation.

https://arxiv.org/abs/2505.09568
Code: https://github.com/JiuhaiChen/BLIP3o
Models: https://huggingface.co/BLIP3o/BLIP3o-Model

7. Skywork-VL Reward by @Skywork_ai

Provides multimodal reward signals using a preference-aligned model that improves mixed optimization techniques and sets new standards on VL-RewardBench.

https://arxiv.org/abs/2505.07263
Model: https://huggingface.co/Skywork/Skywork-VL-Reward-7B

8. Behind Maya

Supports cultural and linguistic diversity in VLMs with multilingual image-text pretraining across eight languages to improve low-resource language understanding.

https://arxiv.org/abs/2505.08910
Code: https://github.com/nahidalam/maya 

9. MiMo by Xiaomi

Unlocks structured reasoning by combining large-scale pretraining with reinforcement post-training on math and coding, beating much larger models on reasoning tasks.

https://arxiv.org/abs/2505.07608
The model checkpoints: https://github.com/xiaomimimo/MiMo 

10. AM-Thinking-v1 by Beike

Elevates open-source reasoning to the 32B scale by fine-tuning Qwen2.5 using RL and SFT, resulting in state-of-the-art scores in math and code.

https://arxiv.org/abs/2505.08311 

Read other interesting AI/ML news of the week, as well as freshest research, in our free newsletter: https://www.turingpost.com/p/fod101