Skip to main content
ModelDescriptionTypeStateCreatorHugging Face IDWeight dtypeActivation dtypeKV Cache dtypeFine-TuningServerlessContext LengthEmbeddingsInput ModalityOutput Modality
MiniMax M2.5MiniMax M2.5 is a frontier mixture-of-experts model with 230B total / 10B active params, trained via large-scale reinforcement learning. Scores 80.2% on SWE-Bench Verified and 76.3% on BrowseComp.LLMReadyMiniMaxMiniMax-M2.5FP8FP16FP16Contact SalesSupported197KNot SupportedTextText
Kimi K2.5Kimi K2.5 is a native multimodal MoE with 1T total / 32B active params, supporting instant and thinking modes, agent swarm coordination of up to 100 sub-agents.LLMReadyMoonshot AIKimi-K2.5BF16BF16BF16Contact SalesSupported262KNot SupportedText / ImageText
GLM 5GLM-5 is a MoE model from Z.ai with 744B total / 40B active params, trained on 28.5T tokens. Scores 73.3% on SWE-bench Verified.LLMContact SalesZ.ai (Zhipu)GLM-5BF16BF16BF16Contact SalesSupported203KNot SupportedTextText
DeepSeek V3.2DeepSeek V3.2 is a MoE with 671B total / 37B active params. Introduces ‘thinking with tools’ capability; achieves gold-medal results at 2025 IMO and IOI.LLMReadyDeepSeekDeepSeek-V3.2FP8FP16FP16Contact SalesSupported164KNot SupportedTextText
gpt-oss-120bOpen-weight MoE from OpenAI with 117B total / 5.1B active params per forward pass, runs on a single H100 or AMD MI300X GPU using MXFP4 quantization.LLMReadyOpenAIgpt-oss-120bBF16BF16BF16Contact SalesSupported131KNot SupportedTextText
gpt-oss-20bOpen-weight MoE from OpenAI with 21B total / 3.6B active params, designed for low-latency inference on consumer or single-GPU hardware.LLMContact SalesOpenAIgpt-oss-20bBF16BF16BF16Contact SalesSupported131KNot SupportedTextText
Qwen3 InstructMoE LLM with 235B total / 22B active params, optimized for instruction following across 100+ languages. 256K token context window.LLMReadyAlibaba / QwenQwen3-235B-A22BBF16BF16BF16Contact SalesSupported256KNot SupportedTextText
Qwen3 ThinkingMoE reasoning model with 235B total / 22B active params, designed for deep logic, math, science, and complex multi-step coding tasks.LLMReadyAlibaba / QwenQwen3-235B-A22BBF16BF16BF16Contact SalesSupported256KNot SupportedTextText
Qwen3 CoderMoE model with 480B total / 35B active params, purpose-built for agentic coding. Native 256K context window, scalable to 1M tokens.LLMReadyAlibaba / QwenQwen3-Coder-480B-A35B-InstructBF16BF16BF16Contact SalesSupported262KNot SupportedTextText
Qwen3.5Multimodal MoE with 397B total / 17B active params and hybrid Gated Delta Network architecture enabling 1M token context window.LLMReadyAlibaba / QwenQwen3.5-397B-A17BBF16BF16BF16Contact SalesSupported1MNot SupportedText / ImageText
Qwen3 VL InstructVision-language MoE with 30B total / 3B active params, supporting megapixel-level inputs, multilingual OCR, visual grounding, and GUI automation.LLMReadyAlibaba / QwenQwen3-VL-235B-A22B-InstructBF16BF16BF16Contact SalesSupported256KNot SupportedText / ImageText
Qwen3 ASRAdvanced speech recognition with 1.7B params, supporting 52 languages, streaming and offline inference. Built on Qwen3-Omni architecture.AudioReadyAlibaba / QwenQwen3-ASR / Qwen3-TTSBF16BF16BF16Contact SalesSupportedN/ANot SupportedAudioText
Qwen-ImageOpen-weight diffusion transformer from Alibaba with strong text rendering accuracy and multilingual typography.ImageReadyAlibaba / QwenQwen-Image-2512BF16BF16N/AContact SalesSupportedN/ANot SupportedTextImage
Qwen-Image-EditDedicated image editing variant of Qwen-Image series, supporting style transfer, object insertion, and detail enhancement.ImageReadyAlibaba / QwenQwen-Image-Edit-2511BF16BF16N/AContact SalesSupportedN/ANot SupportedText / ImageImage
Flux2Production-grade image generation from Black Forest Labs with 9B parameter flow transformer, using rectified flow matching for efficient inference.ImageReadyBlack Forest LabsFLUX.2-Klein-9B / FLUX.2-Klien-4BBF16BF16N/AContact SalesSupportedN/ANot SupportedTextImage
Stable Diffusion 3.5Diffusion transformer from Stability AI with the most mature open-source tooling ecosystem (ComfyUI, Automatic1111, Forge).ImageContact SalesStability AIstable-diffusion-3.5-largeBF16BF16N/AContact SalesSupportedN/ANot SupportedTextImage
Hunyuan ImageLarge-scale multimodal autoregressive image generation from Tencent with 80B total params across 64 MoE experts, trained on 5B image-text pairs.ImageReadyTencentHunyuanImage-3.0BF16BF16N/AContact SalesSupportedN/ANot SupportedTextImage
Z-ImageOpen-weight diffusion transformer (Apache 2.0) for ultra-fast inference with accurate bilingual text rendering in English and Chinese.ImageContact SalesTongyi-MAI / AlibabaZ-Image-TurboBF16BF16N/AContact SalesSupportedN/ANot SupportedTextImage
Wan2.2-I2VOpen-source video diffusion model from Alibaba for image-to-video generation with strong motion consistency and scene coherence.VideoReadyAlibaba / WanWan2.2-I2VFP16FP16N/AContact SalesSupportedN/ANot SupportedText / ImageVideo
Wan2.2-T2VOpen-source video diffusion model from Alibaba for text-to-video generation with improved motion dynamics over Wan2.1.VideoReadyAlibaba / WanWan2.2-T2VFP16FP16N/AContact SalesSupportedN/ANot SupportedTextVideo