Models & assistants Model or AI service
Thinking Machines releases Inkling 2 Mini with open weights
Thinking Machines Lab released Inkling 2 Mini, a multimodal model accepting text and images and generating text, with open weights and hosted access options. [1][2]
What the model does
The model reuses selected transformer layers during inference. Its official card lists Apache 2.0 licensing and local deployment support on specified hardware, alongside access through the Tinker platform. [1][2]
Why it matters
Developers can assess a smaller model for local assistants, coding and tool-use workflows, while researchers can examine the repeated-layer design. [1][2]
Key dates
Model limitations
Local operation requires suitable memory and an inference framework. The card warns about incorrect facts, inconsistent instruction following and weaker long conversations; reported benchmark advantages do not guarantee results on individual workloads. [1][2]
Sources and verification
Catalog evidence checked How we verify records
- Primary source Inkling2Mini model card
- Primary source Introducing Inkling2Mini
Related profiles
Models & assistants
Ayb 0.5 Armenian model family
Metric AI
The release adapts the Qwen3.5 architecture through Armenian tokenizer expansion and language training, while retaining its vision components. The publisher links model repositories and an evaluation collection.
Models & assistants
GPT-Live-1.1 API
OpenAI
The API update adds a model option for multilingual voice conversations and supports delegation to a backend model. A standby configuration for WebRTC sessions pauses inference during inactivity and resumes it when speech returns.
Models & assistants
Gemini 4 Neon
Google DeepMind
The model supports a one-million-token context window. Google announced rollout through the Gemini API and AI Studio, Gemini for business, and eligible Pro and Ultra access in the Gemini app and Antigravity.