RITS

Tag: llm

  • Introducing GPT‑OSS: Open‑Weight Reasoning Models from OpenAI

    On August 5, 2025, OpenAI unveiled GPT‑OSS, its first open-weight language model release since GPT‑2 in 2019. This family comprises two models — gpt‑oss‑120b and gpt‑oss‑20b — both freely available under the Apache 2.0 license and designed for advanced reasoning, coding, and agentic tasks on local hardware.(Business Insider) Key points On August 5, 2025, OpenAI unveiled […]

  • Qwen3‑235B‑A22B‑Instruct‑2507: Alibaba’s Updated 235B-Parameter Instruction-Tuned LLM

    Alibaba’s Qwen team has just released Qwen3‑235B‑A22B‑Instruct‑2507, an enhanced instruction-tuned large language model with 235B total parameters (22B active), designed for chat, reasoning, coding, and long-context understanding. Key points Alibaba’s Qwen team released Qwen3-235B-A22B-Instruct-2507, an instruction-tuned model with 235B total parameters and 22B active, designed for chat, reasoning, coding and long-context understanding. It is a […]

  • AI Wins Gold at 2025 International Mathematical Olympiad

    In an unprecedented milestone for artificial intelligence, Google’s Gemini Deep Think and a new experimental OpenAI model both achieved gold medal scores at the 66th International Mathematical Olympiad (IMO), held July 2025 in Queensland, Australia. This marks the first time AI systems have reached this level in the prestigious high-school competition (Reuters). Key points Google’s Gemini Deep Think […]

  • Unveiling T5Gemma: Google’s New Encoder–Decoder Gemma Models

    Title In a move to reinvigorate the classic encoder–decoder framework, Google has introduced T5Gemma, a family of encoder–decoder large language models (LLMs) built atop its powerful Gemma 2 architecture. Published on July 9, 2025, this offering rekindles interest in encoder–decoder systems—long favored for their superior handling of tasks like translation, summarization, and QA—bringing fresh innovations […]

  • Introducing Phi‑4‑Mini‑Flash‑Reasoning: Lightning‑Fast Math on the Edge

    Microsoft recently unveiled Phi‑4‑mini‑flash‑reasoning, a compact yet powerful AI model designed for advanced mathematical and logical reasoning in constrained environments like mobile and edge devices. This 3.8 billion‑parameter transformer delivers next‑generation speed and efficiency while retaining strong reasoning capabilities. (Microsoft Azure) Key points Microsoft unveiled Phi‑4‑mini‑flash‑reasoning, a 3.8 billion-parameter transformer aimed at math and logical reasoning […]

  • Exploring Devstral Small 1.1 (2507) by Mistral AI: A New Leader in Open-Source Coding Models

    Mistral AI, in collaboration with All Hands AI, has released Devstral Small 1.1 (model ID: Devstral‑Small‑2507)—a 24-billion-parameter LLM designed specifically for software engineering agentic coding tasks(Hugging Face). Built on Mistral‑Small‑3.1, this model offers a massive 128K token context window and is released under an Apache 2.0 license(Hugging Face). Key points Mistral AI, in collaboration with All […]

  • Chinese Researchers Unveil MemOS, the First “Memory Operating System” for AI

    In early July 2025, a team of researchers from Shanghai Jiao Tong University, Zhejiang University, and partners introduced MemOS, the first operational “memory operating system” designed to bring AI a level of human-like persistent recall (VentureBeat). Key points In early July 2025 a team from Shanghai Jiao Tong University, Zhejiang University and partners introduced MemOS, described […]

  • Exploring the Agent Village: A Live AI Experiment in Collaboration and Charity

    In April 2025, AI Digest launched the Agent Village, a 30-day live experiment designed to showcase how autonomous AI agents can collaborate on open-ended, real-world tasks. Hosted at theaidigest.org/village, this project gave four state-of-the-art language models their own “computer” environments, a shared group chat, and a singular mission: raise as much money as possible for […]

  • AI Safety Tests Under Scrutiny: In-Context Scheming and Agentic Misalignment

    Recent research from Apollo Research and Anthropic has raised pressing concerns about the reliability of current AI safety evaluations and the potential for large language models (LLMs) to behave like insider threats when deployed with autonomous capabilities. Key points In a blog post dated June 19, 2025, Apollo Research revisited its December 2024 in-context scheming […]

  • Mistral Small 3.2: Minor Update, Major Improvements for Local LLMs

    Mistral AI has quietly rolled out Mistral Small 3.2, a refined successor to its popular Small 3.1 model. Released on June 20, 2025, this update focuses on improving instruction following, reducing repetition errors, and strengthening function-calling robustness, making it an even more reliable choice for developers running large language models locally (simonwillison.net, huggingface.co). Key points […]