AI Platforms, get the model running

ByteDance flagship released 2026/6/23, rivaling GPT-5.5/Opus 4.7; 10+ hour autonomous engineering, 2-hour video understanding, 80% lower cost than Opus

Meta x EssilorLuxottica AI smart glasses from $299, with native Muse Spark multimodal model; 26 styles, targeting 10M units in 2026 H2

Mistral AI optical character recognition model, extracting text from documents and images

OpenAI first self-developed AI inference chip with Broadcom, TSMC 3nm, 9-month tape-out; ~50% lower inference cost than mainstream GPUs

China market regulator 2026/6/26 released 7 national standards for AI agent interconnection, covering architecture, identity, description, discovery, interaction, and tools

Tencent Yingyongbao OS-level AI assistant released 5/21, Win/Mac/Android, 6 agents, on-device privacy + cross-device, 10M free tokens daily

Cohere open-source model released 5/22, 218B MoE, Apache 2.0, single B200, 48 languages, native citation

Anthropic enterprise AI security architecture launched 5/23, multi-layer sandbox + explainable audit, full decision trail for external calls

Google I/O 2026 24/7 personal AI agent, based on Gemini 3.5 Flash + Antigravity, background continuous work for daily tasks

Leading LLM app framework with chain calling and Agent capabilities

Open-source local LLM runner, one command startup, community favorite

Ultra-fast AI inference with proprietary LPU chip, developer favorite

AI model community by Alibaba DAMO, 300+ models, China's HuggingFace

Open-source all-in-one AI client, multi-model, community favorite

AI API aggregation: one API for 400+ models, developer essential

DeepSeek new-generation flagship model with improved reasoning, coding, and multilingual capabilities

Meituan's new large model preview launched on 2026/9/25, continuing the MoE architecture with native image understanding and deep adaptation to Claude Code

NVIDIA multimodal AI agent model, efficient inference for edge and robotics

FaceBrain open-source multimodal model, vision/audio/text, efficient on-device deployment

2026's breakout general AI agent, autonomous complex tasks, 22M monthly visits

Ant Group large language model for enterprise and financial scenarios

Xiaomi open AI model platform for developers, supporting various MiMo models

AI agent control panel for managing and orchestrating multiple AI agents

Huawei Cloud agent development platform for building and deploying enterprise AI agents

Apple Intelligence 2026 with Siri powered by Gemini integration, enhanced on-device AI capabilities

AI vector database and embedding tool for semantic search and RAG applications

NVIDIA spatial intelligence tool for 3D understanding and robotic spatial reasoning

SDK for building and controlling robot agents with AI capabilities

NVIDIA comprehensive safety system for robotics, ensuring safe AI agent operation in physical environments

DeepSeek's smallest new-architecture model, launched Sep 10, 2026 and refreshed to the 1017 checkpoint on Oct 17: 552B backbone plus 196B Engram (about 763B total weights), 8B/16B active params, native vision, 1M context, FP4 KV cache at ~890 bytes/token; the 1017 build scores 77.0 on DeepSWE v1.1, MIT-licensed with API input from 1 yuan per million tokens

Sakana AI orchestration model that routes across a swappable pool of open and specialized models; Fugu Max is priced at $2/$6 per million tokens

Xiaomi's native omni-modal open-weight flagship, a sparse MoE with 1.02T total and 42B active parameters, 1M-token context, top open-weight score of 46 on the Artificial Analysis Intelligence Index, MIT licensed

NVIDIA's open-weight hybrid model with 30B total and roughly 3B active parameters, up to 1M-token context and tool calling, built for high-frequency agent execution steps and local deployment

Apache-2.0 OS-level agent sandbox using Seatbelt and bubblewrap isolation

Microsoft multi-agent coding harness, no orchestrator, scales to 1024 agents

LlamaIndex doc parsing: PDFs and tables into Markdown, powered by OCRia-1

High-performance open model inference with training and fine-tuning

Anthropic API platform with top-tier Claude model interfaces

GenAI computing infra, developer-friendly, cost-effective inference

No-code AI app platform by Baidu, drag-and-drop smart app builder

AI no-code app platform for rapidly building and deploying AI apps

Xiaomi MiMo text-to-speech model with natural voice synthesis

Agent community platform for discovering and sharing AI agents

Anthropic enterprise AI security model, vulnerability detection, threat intelligence, malware analysis, red-blue teaming

Flagship model released by Anthropic on 2026/9/1 (public version sharing the same foundation as Mythos 5.1), officially positioned as the world's most advanced model for programming and knowledge work: Terminal-Bench-Science 0.1 52.6%, CursorBench 73.4%, Terminal-Bench 4.0 55.8%; cache read price reduced by 75% to $0.25, typical load cost reduced by about 25%; 1M context/128K output

Anthropic's research-restricted flagship released on 2026/9/1 (sharing the same underlying model as Fable 5.1, with relaxed guardrails in dual-use domains), available only to cybersecurity and life sciences institutions reviewed under the Trusted Access Program; its network capabilities surpass the previous generation Mythos 5, with about half of 12 protein design targets hitting the mark; its capabilities also support Claude Security for enterprise customers

NVIDIA open-source 4B world model for robots and edge AI, real-time environment understanding and action generation

Xiaohongshu LLM series, dots-note 3.0 scored perfect 42/42 at IMO 2026; reads raw LaTeX end-to-end without formal languages

FaceBrain + OpenBMB open-source embodied AI model series, 1.5B VLA manipulation + 0.9B tracking

Provides AI Agent with real-time, authoritative, and reliable internet search capabilities, supporting cross-language, multimodal, and multi-domain information retrieval

A new paradigm for multimodal large model training open-sourced by Xiaohongshu, achieving efficient training through a dependency-safe nested pipeline

LLM plugin, starts a local server compatible with the OpenAI Chat Completions API, exposing all installed models.

Open-source runtime providing virtual file systems and multi-execution environments for AI agents

Cloud-native computing platform supporting inbound TCP connections and gRPC

High-performance inference engine optimized for local agent workflows, supporting Day-0 deployment of multimodal models such as Muse Glimmer.

AI extension integrating Alibaba's Qwen model on Apple Mac, supporting writing tools and Siri intelligent Q&A.

An open-source unified agent plugin specification supported by Google, Amazon, Microsoft, etc., packaging Agent Skills and MCP servers as portable units.

HPC operator library open-sourced by Tencent Hunyuan, integrated into SGLang, providing high-performance Attention, Router GEMM, and MoE operators, reducing TPOT by up to 48.8%.

AI teammate app launched by xAI in collaboration with Cursor. The Bot has an independent virtual computer, can log into external services to execute multi-step tasks, and supports Mac, Windows, and iPhone

Meta's open-source 30B dense multimodal model, under Apache 2.0 license, designed for local agent scenarios, runs on a single consumer-grade GPU, supports image, text, and video input with tool calling

Formerly LambdaTest's Agentic AI quality engineering platform, Kane AI automatically plans and writes tests, covering real device cloud, cloud execution, and AI Agent testing

API infrastructure providing dedicated mailboxes for AI agents. Agents can send and receive emails, manage conversation threads via API and SDK, and support MCP integration.

Six fully open model family (0.9B–375B-A23B) released by IFM (under MBZUAI) at the Abu Dhabi Foundation Model Institute on 2026/9/3, licensed under Apache 2.0 with public weights, training code, data ratios, and intermediate checkpoints; the flagship 375B-A23B activates about 23B per token, natively supports 512K context, and scores 70.2% on Terminal-Bench 2.1

Yutori released on 2026/8/26 a 27B general-purpose computer-use model: supports full desktop operations (Linux/macOS/Windows), no longer limited to browsers; OSWorld 2.0 65.2% (about $1.46/task), OSWorld-Verified 85.3%, MacAgentBench 83.1%; API input $0.5/million tokens, output $4

Berkeley RDI's open-source platform for computer-use agents: Lite.Gym, Lite.Sample and per-model harnesses unify environments, data and training pipelines; 15+ benchmarks plug in directly, VM-free desktop sandboxes ship 30k+ verifiable tasks, and 14 model families are supported for evaluation and training

Reflection's first open-weight model (Oct 16, 2026): 501B MoE with 23B active, 1M-token context, text-only, Apache 2.0 weights in BF16/FP8/NVFP4, built for coding and agentic work; open-source Mirror CLI and 40+ day-0 partners, founded by ex-DeepMind researchers

OpenAI's cloud agent API in public beta (Sep 10, 2026): the harness and infrastructure that power Codex, invoked with a single API call, with managed sandboxes, self-hosted options, context compaction, tool search and parallel subagents; no platform fee, pay per token and tool usage

Search-agent model family open-sourced by Xiaohongshu's AllSpark team on Sep 14, 2026: Iris-mini (35B, built on Qwen3.6-35B-A3B) and Iris-pro (397B, built on Qwen3.5-397B-A17B), both with 256K context; leads open models of comparable size across BrowseComp, BrowseComp-ZH, DeepSearchQA and HLE, with Iris-mini's 82.2 on BrowseComp approaching trillion-parameter models; trained with an SFT-to-RL climbing recipe, in-cluster judging and summarization, and resumable ultra-long trajectories; weights, paper and model cards are public

Scientific multimodal foundation model from Shanghai AI Laboratory, announced Sep 13, 2026 at the Pujiang Innovation Forum and released as a full open-source build on Hugging Face on Sep 14: 397B parameters focused on long-horizon research and agentic ability, with general knowledge, coding and agent skills in the top tier of open models; a pluggable Memory Decoder adds domain knowledge without changing base weights; scores 87.00 on FrontierScience-Olympiad, 93.56 on HMMT-2026 and 68.54 on SWE-bench-Pro; deeply co-optimized with the Ascend compute ecosystem and set to plug into the Shusheng Duanyan scientific discovery platform

Natively omnimodal model released by Alibaba Qwen on 2026/9/17 with text, image, audio and video input and a 1M token context; Alibaba reports the average score across 29 benchmarks is more than 25% above Qwen3.5-Omni-Plus while hourly audio input pricing falls over 98% and audio-video input over 93%, aimed at audio and video agent delivery

Open-source fine-tuning and training acceleration toolkit; on 2026/9/17 it shipped a Docker image and the Unsloth Desktop GUI so more than 500 models can be trained and run locally without configuration, supporting both NVIDIA and AMD graphics cards with GUI and notebook workflows

Lightweight open-source translation model Google released on 2026/9/15, trained from Gemini, supporting 55 languages and able to run offline; Google says its language technology now covers more than 300 languages reaching roughly 86% of the world population

Qwen's real-time speech interpretation model released 2026/9/18, built on a single-stream Interleave architecture with a Hybrid-MoE Thinker-Talker design, cutting average interpretation lag (LAAL) from 2.8 to 2.3 seconds; accepts text translation input in 60 languages with spoken output in 29, with built-in real-time speaker diarization, bilingual source-translation alignment and long-context disambiguation; served via Alibaba Cloud Bailian's realtime API

TypeSafe AI's large model built for high-frequency decisions: no dialogue or prose generation, only typed decisions with calibrated confidence, trained with the company's RLCD method; by its own workflow benchmarks up to 193.6x faster and 444.6x cheaper than conventional large models; access by application on the official site, with Vercel as the launch integration

The efficiency tier of the MiMo-V2.6 series, a sparse MoE with 309B total and 15B active parameters, 1M-token context, native omni-modal, released and open-sourced alongside Pro under MIT

The high-speed serving tier of Xiaomi MiMo-V2.6-Pro, sharing the same checkpoint while advertising up to 20x output speed at roughly ten times the standard price, built for real-time interaction and latency-sensitive production workloads

A family of general-purpose computer-use agent models released by H Company on September 28, 2026, including two sizes, 27B and 35B-A3B, capable of completing tasks across desktop, web, mobile, and API.

An open-source agent safety platform released by NVIDIA on September 27, 2026, which prevents AI agent behavior drift and privilege escalation through kernel-level isolation, a zero-trust environment, and out-of-band monitoring.

An open-source large model inference engine that uses PagedAttention to manage key-value caches and provides a high-throughput, OpenAI-compatible server. In September 2026, lossless text watermarking was added.

A MoE large model released by Ant Bailing (InclusionAI) on September 30, 2026, with approximately 560B total parameters and about 25B activated per token, and a maximum context window of 1M; continuously optimized around general-purpose agents, search, daily office work, and software development, with enhanced capabilities in scenarios such as healthcare, finance, and materials science; a two-week free trial is available (the service length during the trial period is 256K), after which it becomes paid and is planned to be open-sourced

Prime Intellect inference platform: serverless endpoints and reserved capacity, automatic failover across data centers, OpenAI-compatible API, launched with GLM-5.3

A decision model from Liquid AI that takes text or images plus questions and returns a probability for each answer in one forward pass without generating tokens, with text decisions in 200 to 300 ms and billing on input tokens only

Mistral released this frontier model in Research Public Preview on October 6, 2026. It has 1T total parameters with 49B active, native multimodality, a 512K context window and up to 256K output tokens. It scores 38 on the Artificial Analysis Intelligence Index, level with GPT-6 Luna at max effort and the highest score from outside the US and China, plus 50 on the Cyber Index, with 1T weights planned for release at the end of October

Google DeepMind released this multimodal embedding model in October 2026. Built on the Gemma 4 architecture with 740M parameters and open weights under Apache 2.0, it maps text, code, images, video frames and audio into a single embedding space, so cross-modal queries such as searching video with words work directly, at a size that still fits on-device and in-browser deployment

Sierra and Meta announced this open standard on October 6, 2026 with partners including Genesys, Instinct, Rocket, Shopify, Stripe and Walmart, defining how personal AI agents interact with businesses. An agent first discovers on the company website what is offered and how to reach it, then opens a session on the user behalf over OAuth, starting as a guest for things like stock checks and letting the user decide between read-only and write access once an account is involved, and completes tasks through the company website, its MCP or OpenAPI based APIs, or the company own agent. The v0.1 specification and a reference implementation are planned for later this month

The Google AI Edge team open-sourced this cross-platform on-device GPU compute engine under Apache 2.0 on October 8, 2026. Built for on-device AI inference, it serves as the core GPU acceleration layer inside LiteRT while also standing alone as a library. It abstracts OpenGL ES, OpenCL, Metal and WebGPU behind one API, uses tensor virtualization to decouple logical tensors from physical GPU allocation so a single shader codebase covers every backend, adds 5D tensor support and an extensible custom op framework, and switches kernels and layouts between the prefill and decode phases of on-device LLMs. It already runs daily across millions of devices in Chrome, YouTube Shorts, Google Photos and Meet

Google open-sourced this ambient quality agent on October 8, 2026. It runs unattended inside your own Google Cloud project beside your agent, sweeping production trajectories from Cloud Trace, Cloud Logging or BigQuery on a schedule, after each deployment or on demand. A five stage pipeline samples up to 1,000 recent sessions, grades each against a nine point checklist of common breakdowns, clusters findings that share a failure mechanism, has a separate model verify each cluster against up to three full transcripts, then tracks surviving clusters in BigQuery as new, recurring or auto-resolved after 14 days unseen. Raw transcripts, source snapshots and tables stay inside your project boundary, and AQuA never sits in the request path or writes back. Root cause analysis cites line ranges against the deploy-time source snapshot and proposes an edit, but never applies one or opens a pull request on its own

Microsoft released this decision model on October 9, 2026. Post-trained from Qwen3.5-9B, it turns text and images into structured answers such as probabilities, options and scores, and is available on Microsoft Foundry and OpenRouter. Microsoft says it records the highest accuracy across about 36 blind benchmarks covering roughly 150,000 questions and runs 4.5x faster than the runner-up, at $0.042 per million input tokens with free output and a 32K context.

Cloudflare released this open-weight multimodal decision model on October 9, 2026. A 30B-total / 3B-active MoE post-trained from Qwen3-Omni-30B-A3B-Instruct, it scores text, images, audio and video in a single call and returns a calibrated probability for every option rather than generating text, supporting true/false, multiple choice and ordinal score questions. Weights ship under Apache-2.0; on Workers AI it runs as @cf/cloudflare/clef-omni with a 64K context, about 130 ms median latency for text-only decisions, $0.15 per million input tokens and no output charge.

Nace AI open-sourced this 8.95B dense decision model. It reads a text or JSON state plus typed questions and returns a probability for every option without generating any text, supporting choice, yes/no and ordinal score questions over a POST /v1/systemone endpoint. It scores 58.08 on the public Decision Index 0.3.1, inside the tie band of the closed-source Jev 1.13.0 and the highest result for any model under 10B parameters, with 93.4% accuracy on 32K to 128K token documents at 2.0 s median latency. Weights are on Hugging Face, a 9.5 GB Q8_0 variant runs on Apple silicon and CPU, and hosting on OpenRouter costs $0.04 per million input tokens.

Pine AI launched this developer-facing cloud computer on October 9, 2026. Through an SDK, a product creates a Pine Computer whenever a job needs one and hands over the task, and the agent works directly across websites, files and business software before returning finished output. The company reports a 78.3% checkpoint score on UniPat AI SaaS-Bench v1.1. It is in private beta, with a waitlist at pinecomputer.io.

Meta flagship model line released on 2026/10/12. Muse Wonder and Muse Wonder Full score 52 and 56 on the Artificial Analysis Intelligence Index, with 1M context, built for personal and professional agent use.

Open-weight model from Liquid AI released on 2026/10/13, 24B total parameters with 2B active per token, 131K context, reasoning first, built for fast private agents on a laptop or your own hardware

Robotics foundation model API opened in beta by Physical Intelligence on 2026/10/13: upload data, fine-tune the pi model family, and run inference on your own robots through a hosted gateway

An open-source general-purpose computer-use agent technology stack from Alibaba Tongyi MAI-UI team, including harness, cross-device SDK, macOS app, and evaluation trajectories

An open-source native PyTorch backend from AWS that runs models directly on Trainium chips through the neuron device, supporting eager execution and torch.compile

Proactive agent on Opus 5.5: briefs, plus tasks run in its own cloud

OpenAI endpoint returning probabilities, choices and scores

30B-A4B MoE trained from scratch in Korea, 256K context, Apache 2.0

Liquid AI's 2026/10/27 release of six open-weight 2B-class models, all branched from one tri-hybrid backbone and covering text, vision, audio, omni and decision work, with the text model at 1.73B parameters, a 256K context and about 1GB of memory at 4-bit

Open-source visual LLM orchestration tool, drag-and-drop RAG/Agent/chat apps, 35K+ GitHub stars

LangChain visual builder, drag-and-drop LLM apps and Agent flows, 45K+ GitHub stars

OpenAI API platform with GPT, DALL·E, Whisper model interfaces

World's largest AI open-source community and model hub: 18M+ developers, 3M+ models, 500K+ datasets; acquired by NVIDIA for $12.93B in 2026/9

Google's AI dev platform for using Gemini series models

AI Bot development platform by ByteDance for creating AI apps

Open-source LLM app platform with visual AI workflow orchestration

Google open-source multimodal model family April 2026, 2.3B to large, strong reasoning and Agent ability

Meta multimodal model for smart glasses and wearables, powering Meta Glasses 2026

Desktop app for running local LLMs, download and chat with open-source models offline

Open-source web interface for running local LLMs, Ollama-compatible, extensible with plugins

AWS managed service for foundation models, accessing Claude, Llama, Titan via unified API

Google Cloud unified ML platform, model training, deployment, and GenAI application building

Microsoft Azure hosted OpenAI models with enterprise security, regional availability, and compliance

Fast inference platform for open-source LLMs, low-latency API for production deployments

Serverless model deployment platform, fast inference for open-source and custom models

Serverless cloud platform for AI/ML workloads, auto-scaling Python code execution

SenseTime daily-refresh model (RiRiXin) for enterprise multimodal AI applications

AI automation tool for desktop and web task execution

Cloud AI model platform, run open-source models with one line of code

Baidu's large model platform with Wenxin series model APIs

Alibaba Cloud's model service supporting Tongyi and open-source models