Since Claude Opus 4.6, writing quality has declined with a distinct "Claude style." An Anthropic engineer explained that later versions prioritized math reasoning and coding to boost complex engineering tasks, sacrificing natural writing and revealing a side effect of model evolution.....
Microsoft reportedly plans to cut enterprise Copilot subscription prices by up to 50%, possibly in October, and launch a "super AI app" to evolve it from office assistant to automation platform, using price cuts and upgrades amid fierce AI assistant competition.....
On September 22, the Yunchi Conference in Hangzhou opened, and Alibaba announced the progress of large models: Qwen4 has entered training, and the subsequent parameter scale is planned to expand to 5 trillion to 10 trillion; recursive self-improvement enters training, reasoning, and chip coordination; multimodal matrix upgrade, a new video generation model will be released in November. Liu Dayi Heng said that scaling is an important path toward ASI.
OpenAI is building AI to control browsers and apps, evolving ChatGPT from a generative tool into a digital assistant that operates software. It reportedly targets SpaceX's GrokBot, enabling always-online AI partners to handle daily tasks, manage software, and work in the background—a tireless digital partner.....
Knowledge base-based white-label AI customer service that delivers instant replies, captures leads, books appointments, and hands off to human agents.
Personal AGI with an original kernel, supporting multimodality, built - in 54 skills, and can run locally.
Apodex is a self-evolving heavy-duty solver that conducts in-depth research on difficult problems and verifies the results of each step.
Cola is an Agent OS and AI operating system. It comes pre - installed with model tools and can evolve together with humans.
thenexthub
OpenModel-1T-A50B-Instruct is a trillion-parameter mixture-of-experts model developed by NeXTHub. It combines the evolutionary thought chain training method to achieve a balance between reasoning ability and energy efficiency, and has deep reasoning ability and 128K long-context processing ability.
andrewdalpino
The ESMC Protein Function Predictor is a tool based on the Evolutionary Scale Model (ESM) that can predict protein functions using the Gene Ontology (GO) based on amino acid sequences. This model is pre-trained on the UniRef, MGnify, and Joint Genome Institute databases and fine-tuned on the AmiGO Boost dataset. It can predict the molecular functions, biological processes, and cellular locations of proteins.
A protein function prediction tool based on the Evolutionary Scale Modeling (ESM), which predicts the molecular function of proteins through amino acid sequences.
alvdansen
The Kodak model perfectly replicates the nostalgic essence of early 1990s photography, evoking memories of disposable cameras and carefree travels.
SakanaAI
Llama-3-EvoVLM-JP-v2 is an experimental general-purpose Japanese vision-language model that supports interleaved input of text and images. This model was created using an evolutionary model fusion approach.
Experimental Japanese Ukiyo-e style image generation model for educational purposes
umiyuki
Umievo-itr012-Gleipnir-7B is a 7B-parameter language model obtained by fusing four powerful Japanese models using evolutionary algorithms, excelling in Japanese language tasks.
allganize
Alpha-Instruct is an outstanding language model in Korean and English domains, developed using evolutionary model fusion technology, particularly excelling at complex linguistic tasks and logical reasoning.
kuleshov-group
PlantCaduceus is a DNA language model pre-trained on 16 angiosperm genomes, utilizing Caduceus and Mamba architectures to learn evolutionary conservation and DNA sequence syntax through masked language modeling objectives.
PlantCaduceus is a DNA language model pre-trained on the genomes of 16 angiosperm species, utilizing Caduceus and Mamba architectures to learn evolutionary conservation and DNA sequence syntax through masked language modeling objectives.
freewheelin
free-solar-evo-v0.13 is a model developed by the Freewheelin Artificial Intelligence Technology Team. It draws on the evolutionary model fusion concept of the Sakana project and is developed based on the free-solar-evo-model, with certain technical foundation and application value.
piotrnobis
A deep learning model for solving the inverse problem of Burgers equation, capable of predicting velocity evolution and estimating physical parameters
EvoVLM-JP-v1-7B is an experimental general-purpose Japanese vision-language model created using evolutionary model fusion methods
togethercomputer
Evo is a biological foundation model capable of long context modeling and design. It uses the StripedHyena architecture and can model sequences at single nucleotide and byte-level resolution.
Evo is a biological foundation model capable of long-context modeling and design, employing a striped hyena architecture that models sequences at single-nucleotide byte-level resolution.
Nondzu
Mistral-7B-codealpaca is a code assistant model optimized based on Mistral-7B-Instruct-v0.1, specifically fine-tuned for programming tasks. This model is trained using the evol-codealpaca-v1 dataset and performs excellently in code generation and understanding, which can serve as a programming assistant for developers to improve development efficiency.
WizardLMTeam
WizardCoder is a large language model for code enhanced by the Evol-Instruct method, specializing in code generation tasks.
WizardCoder-Python-34B-V1.0 is a large language model for code enhanced by the Evol-Instruct method, specializing in Python code generation tasks.
mlabonne
A code generation model fine-tuned on the Evol-Instruct-Python-1k dataset using QLoRA (4-bit precision) based on the CodeLlama-7b-hf model
WizardMath-7B-V1.1 is a large language model trained on Mistral-7B, specializing in mathematical reasoning. It employs the Reinforced Evolutionary Instruction (RLEIF) method to enhance performance, currently achieving SOTA among 7B-scale mathematical models.
A-MEM is a self-evolving memory system designed for coding agents. It uses a Zettelkasten-style knowledge graph structure, can automatically extract, associate, and evolve memories, supports semantic search and graph traversal, and integrates with agents such as Claude Code through the MCP protocol.
Zetmem is an AI-assisted memory server project based on the Zettlekasten method, aiming to provide a persistent context memory system for AI programming tools such as Claude, supporting code memory storage, retrieval, and intelligent evolution across sessions and projects.
The task portal system is a self - evolving general problem - solving institution with core components such as logical reasoning, ethical framework, sequential thinking, and meta - framework, which can safely self - evolve and solve complex problems.
An MCP server that connects Claude and WhatsApp, enabling automated interaction features through the Evolution API
EvolveMCP is an open - source project that enables Claude AI to independently build, install, and use tools, expanding the practical capabilities of AI through Python code execution.
KIP is a graph interaction protocol designed for large language models, connecting LLMs with knowledge graphs and providing capabilities for memory persistence, knowledge evolution, and interpretable interaction.
The PLTM MCP server provides 78 tools for Claude Desktop, conducts AGI experiments based on general physical principles, including functions such as memory operations, diverse retrieval, meta - cognition, and knowledge acquisition, aiming to promote the system's evolution towards higher - order intelligence through principles such as criticality and self - organization.
Agent Farm v3.4 is a system based on AI agent evolution and parallel task execution, which realizes task processing through tool - enhanced agents. The new version introduces the chunked write mode, supports the parallel generation of large documents and code files, improves performance by 8.6 times, and uses local models for result synthesis without relying on cloud tokens.
The MCP Evo Supergateway is a gateway application for connecting to the Evolution API, providing remote connection and tool integration capabilities through the SSE endpoint.
Worker17 is a worker management system that combines 3D monitoring with the MCP protocol. It includes a web visualization interface and a server-side, supporting AI-assisted management through Claude Desktop. The project initially started as a joke and later evolved into a technical exploration of the MCP server.
A Pokémon data MCP server based on PokeAPI, providing functions for querying Pokémon information, move details, and obtaining evolution chains.
An MCP service that automatically optimizes AI prompts based on evolutionary algorithms, iteratively improving prompt performance through genetic algorithms
The NSAF MCP Server is a server-side implementation of the Model Context Protocol (MCP) used to connect AI assistants with the Neural-Symbolic Autonomous Framework (NSAF), providing evolutionary algorithms and architecture comparison functions.
MarecoX MCP Servers is an automated installation script project for deploying three different types of MCP servers: Google Calendar MCP, Evolution API MCP, and Instagram MCP. The project provides two installation methods: one - click installation and step - by - step installation, and supports multiple functions such as event management, message sending, and media management.
This project implements an advanced sequential thinking process based on the Multi - Agent System (MAS), which is built using the Agno framework and provided services by MCP. Compared with simple state - tracking methods, it uses coordinated specialized agents for in - depth analysis and problem decomposition, evolving from a passive thinking recorder to an active thinking processor.
The NSAF MCP Server is a server-side service for the neural-symbolic autonomous framework based on the Model Context Protocol, allowing AI assistants to interact with the NSAF framework through the MCP protocol and providing functions such as running evolutionary algorithms and comparing intelligent agent architectures.
PsiAnimator - MCP is a quantum physics simulation and animation server that integrates the QuTip quantum computing library and the Manim animation engine through the MCP protocol, providing functions such as quantum state management, time evolution, measurement calculation, and high - quality visualization, suitable for quantum mechanics teaching and research.
PokeMCP is a server based on the MCP protocol, which provides detailed Pokémon information query functions, including attributes, moves, evolution chains, etc., through integrating PokeAPI, and supports Pokémon battle simulations.
An MCP server that transforms Claude into a structured interviewer. Through a shared evolution system, each conversation makes the system smarter. The system automatically optimizes interview checkpoints and paths based on Bayesian scoring and community usage patterns, achieving collective knowledge evolution.
Overview: The README.md file provides an overview of the Model Context Protocol (MCP) Server, detailing its components and functions. It explains how MCP enables AI automation through a standardized protocol and connects hosts, clients, servers, local data sources, and remote services. The document also describes the evolution from basic AI chats to advanced AI agents capable of performing automated tasks.