QwQ-32B is Alibaba Qwen's March 2025 open-weights reasoning model, a 32.5-billion-parameter dense transformer fine-tuned from Qwen2.5-32B with supervised fine-tuning and reinforcement learning, positioned against DeepSeek-R1 and OpenAI o1-mini on hard-problem reasoning.
Qwen3-Coder-480B-A35B-Instruct is Alibaba Qwen's May 2025 specialized open-weights coding model, a 480-billion-parameter mixture-of-experts model with 35 billion active per token, 256K native context, and Apache 2.0 licensing.
MiniMax M2 is the 2025 open-weights flagship language model from Shanghai-based MiniMax, a 230-billion-parameter mixture-of-experts model with 10 billion active parameters, positioned around agentic coding and tool use under a modified MIT license.
Kimi K2.5 is Moonshot AI's January 2026 open-weights multimodal flagship, a 1-trillion-parameter mixture-of-experts model with 32 billion active parameters, native image and video understanding, and a 256K context window under a modified MIT license.
Qwen 3.6 is the April 2026 generation of Alibaba's Qwen language and multimodal model family, headlined by the 35B-A3B mixture-of-experts variant with 3 billion active parameters, native 256K context, hybrid thinking mode, and Apache 2.0 open weights.
GLM-5.1 is Z.ai's February 2026 open-weights mixture-of-experts language model, a 754-billion-parameter agentic-engineering specialist with state-of-the-art SWE-Bench Pro performance and an MIT license for commercial use.
Gemma 4 is Google DeepMind's 2026 open-weights model family, spanning dense and mixture-of-experts variants from 2 billion to 31 billion parameters, with native multimodal input, 256K context, and Apache 2.0 licensing.
TML-Interaction-Small is Thinking Machines Lab's May 2026 real-time multimodal interaction model: the first entry in the Interaction Models family, handling concurrent audio, video, and text input and output at human-conversation latency.
Kimi K2.6 is Moonshot AI's April 2026 open-weights frontier model, a one-trillion-parameter mixture-of-experts model focused on agentic tool-use at scale.
GPT-Realtime-2 is OpenAI's May 2026 speech-to-speech realtime voice model, the second generation of the GPT-Realtime API line and the production-grade Realtime API release.
Claude Mythos Preview is Anthropic's April 2026 limited-access model with autonomous vulnerability discovery and exploit-development capabilities.
Zamba2 is a family of open-weights hybrid Mamba2-transformer language models from Zyphra at 1.2B, 2.7B, and 7B sizes
SGLang is a high-performance open-source serving framework for large language models and multimodal models
Rnj-1 (Ramanujan) is an 8-billion-parameter open-weights base and instruction-tuned language model from Essential AI
Runner H is the agentic orchestration product from H Company, dispatching natural-language objectives across browsers and applications by combining a…
MOFGEN is a generative AI system for metal-organic framework discovery developed by CuspAI, combining a language model that proposes MOF compositions with…
Marey is a foundational AI video model developed by Moonvalley and trained exclusively on owned and licensed footage, supporting native HD generation
Holo-1 is an open-weights visual-language model family released by H Company in June 2025 under Apache 2.0
DIAMOND is a diffusion-based world model from the General Intuition founding team that simulates interactive game environments by predicting pixel-level…
Chai-2 is a multimodal generative model for zero-shot antibody and binder design released by Chai Discovery in 2025
Chai-1 is a multimodal foundation model for biomolecular structure prediction released by Chai Discovery in September 2024
Aristotle is a mathematical reasoning model developed by Harmonic that produces formally verified proofs in Lean 4 and reported gold medal level…
AFM-4.5B is a 4.5-billion-parameter dense open-weights language model from Arcee AI, trained on 8 trillion tokens curated by DatologyAI for…
Falcon is the open-weights large language model family developed by the Technology Innovation Institute in Abu Dhabi
Ernie is the family of large language models developed by Baidu, spanning the original 2019 ERNIE pre-trained encoders
Doubao is the family of large language models developed by ByteDance Seed and deployed in the Doubao consumer chatbot app
DeepSeek V3 is a 671-billion-parameter open-weights mixture-of-experts language model released by DeepSeek in December 2024
Yi-Large is a closed-weights large language model developed by 01.AI, released in June 2024 as the lab's proprietary flagship
Phi-4 is Microsoft Research's 14-billion-parameter open-weights language model, released in December 2024 under the MIT license
Phi-3 is Microsoft Research's 2024 small language model family, spanning 3.8B to 14B parameters, trained on curated synthetic and filtered web data under…
Kimi K2 is a 1-trillion-parameter open-weights mixture-of-experts language model released by Moonshot AI in July 2025
Hunyuan is Tencent's family of large language models, spanning a text LLM line that runs from the 389-billion-parameter Hunyuan-Large to the April 2026…
Command R+ is Cohere's 104-billion-parameter enterprise language model, optimized for retrieval-augmented generation with grounded citations
RT-2 is Google DeepMind's vision-language-action model that applies web-scale vision-language pretraining to robotic control
Gemini Robotics is Google DeepMind's vision-language-action model family for physical robots, translating visual observations and natural-language…
Gemini 3 Deep Think is Google DeepMind's enhanced reasoning mode for the Gemini 3 family, spending additional compute on extended chain-of-thought before…
AlphaProteo is Google DeepMind's generative protein-design model that creates novel protein binders for specified molecular targets
AlphaFold 3 is Google DeepMind and Isomorphic Labs' structure-prediction model that extends protein folding to protein-ligand, protein-DNA
GPT-5.4 is OpenAI's March 2026 frontier large language model, adding native computer use, configurable reasoning effort
o4-mini is OpenAI's April 2025 compact reasoning model, designed for fast, cost-efficient inference on tasks that benefit from chain-of-thought…
o3 is OpenAI's April 2025 reasoning model, applying large-scale reinforcement learning to chain-of-thought processes to produce a system that thinks…
GPT-5 is OpenAI's August 2025 multimodal large language model that opened the GPT-5 generation, combining a fast inference model with an integrated…
Claude Sonnet 4.5 is Anthropic's mid-tier model in the Claude 4 generation, released September 2025, combining strong agentic coding performance with…
Claude Haiku 4.5 is Anthropic's fastest production model in the Claude 4 generation, optimized for high-volume agentic workloads, real-time chat
Hermes 4 is an open-weights hybrid-reasoning language model family released by Nous Research in August 2025
Eleven v3 is ElevenLabs's flagship text-to-speech and voice-cloning model, supporting 32 languages with highly expressive
Suno v4 is the fourth-generation music-generation model from Suno, capable of producing complete songs with vocals, instrumentation
Whisper is OpenAI's open-weights automatic speech recognition system, supporting transcription and translation across approximately 100 languages under…
Seedance 2.0 is ByteDance Seed's second-generation text-to-video generation model, distributed through the Doubao consumer app, CapCut
Veo 3 is Google DeepMind's third-generation text-to-video generation model, producing high-resolution video with native audio and tight prompt fidelity
Sora 2 is OpenAI's second-generation text-to-video generation model, producing high-resolution videos up to two minutes long with physics realism and…
Stable Diffusion 3.5 is an open-weights text-to-image model released by Stability AI in October 2024, available in Large and Medium variants and…
Midjourney v7 is the seventh-generation text-to-image model from the bootstrapped San Francisco lab, distributed through Discord and the Midjourney web…
FLUX.2 is Black Forest Labs's second-generation flagship text-to-image model, available in closed-weights commercial and open-weights variants
Imagen 4 is Google DeepMind's fourth-generation text-to-image model, released in 2025 in Standard, Ultra, and Fast variants
DALL-E 3 is OpenAI's third-generation text-to-image model, released in October 2023 and notable for its strong prompt fidelity
gpt-oss is OpenAI's open-weights mixture-of-experts language model family released in August 2025, available in 20B and 117B parameter variants and…
Qwen 3 is the open-weights large language model family released by Alibaba's Qwen team in 2025, introducing a hybrid thinking architecture across a broad…
DeepSeek V4 is a 1.6-trillion-parameter open-weights mixture-of-experts language model released by DeepSeek in April 2026
Mistral Large 2 is a closed-weights frontier language model from Mistral AI, released in July 2024 as the most capable model in the Mistral Large family…
Llama 4 is Meta AI's April 2025 open-weights large language model family, available in Maverick and Scout variants with native multimodal vision…
Grok 4.20 is xAI's 2026 flagship large language model, handling text and images and distributed through X, grok.com, Tesla vehicles, and a developer API
Gemini 3.1 Pro is Google DeepMind's February 2026 flagship multimodal model, handling text, images, audio
Claude Opus 4.7 is Anthropic's April 2026 flagship large language model, combining multimodal text and vision capabilities with hybrid extended-thinking…
GPT-5.5 is OpenAI's April 2026 flagship large language model, combining multimodal text and vision capabilities with optional deep-reasoning variants…
Every notable foundation model from the labs covered on Nextomoro. Filter by modality (text, image, video, audio, multimodal, code) or by lab. Each post covers capabilities, benchmark standing, access, and how the model sits against its peers.
nextomoro tracks progress for AI research labs, models, and what's next.
