Legacy News Archive

Deepseek OCR: Vision‑Language Compression Meets Dynamic OCR

19.11.2025 Announcement deepseek-ocr

Cogito 2.1: A Leap in Efficient, High‑Confidence Reasoning

19.11.2025 Announcement cogito-2.1

Qwen3 VL: Alibaba’s 1 M‑Token Multimodal Engine Powers Vision, Code, and Video

29.10.2025 Announcement qwen3-vl

Deepseek V3.1: 685‑Billion‑Parameter Open‑Source LLM Sets New Benchmark

25.08.2025 Announcement deepseek-v3.1

GPT Oss: Configurable Reasoning and Agentic Efficiency

05.08.2025 Announcement gpt-oss

Pioneering Agentic Code Intelligence

01.08.2025 Announcement qwen3-coder

Gemma3N: Pioneering Elastic Inference and Multimodal Efficiency in Language Models

26.06.2025 Announcement gemma3n

Mistral Small3.2: Enhanced Instruction Handling and Reduced Repetition Errors

20.06.2025 Announcement mistral-small3.2

Magistral: Transparent, Multilingual Reasoning for Global Applications

09.06.2025 Announcement magistral

Agentic LLMs Redefine Software Engineering with Devstral

21.05.2025 Announcement devstral

Advancing Visual Understanding: Qwen2.5Vl's Multimodal Breakthroughs

14.05.2025 Announcement qwen2.5vl

Phi4 Mini Reasoning: Efficient Mathematical Reasoning for Resource-Constrained Environments

01.05.2025 Announcement phi4-mini-reasoning

Microsoft's Phi4 Reasoning Models: Efficient Reasoning in AI

01.05.2025 Announcement phi4-reasoning

Llama4: Pioneering Multimodal AI with Scalable Expertise and Extended Context

01.05.2025 Announcement llama4

Qwen3: Advancing Language Model Versatility with Adaptive Modes

28.04.2025 Announcement qwen3

IBM's Granite3.3: Pioneering Long-Context Language Models with Enhanced Reasoning and Multi-Modal Capabilities

16.04.2025 Announcement granite3.3

Open-Source 14B Model Deepcoder Advances Code Reasoning with 60.6% Accuracy

08.04.2025 Announcement deepcoder

Cogito's Hybrid Reasoning and Self-Reflection Redefine AI Capabilities

08.04.2025 Announcement cogito

Mistral Small3.1: Expanding Context, Enhancing Performance

07.04.2025 Announcement mistral-small3.1

Exaone Deep: Pioneering Reasoning Capabilities in Large Language Models

19.03.2025 Announcement exaone-deep