Inference infrastructure provider serving open and commercial models with a focus on high-throughput GPU-backed endpoints.
60 text
Xiaomi's open-source efficiency model: a 309B MoE with 15B active parameters per token and a hybrid attention mechanism.
Xiaomi's 1T-parameter flagship, built to push the capability ceiling on the most demanding workloads.
InclusionAI's hybrid vision-language model, layering native visual perception on the Ling 3.0 Flash MoE (124B total, 5.5B active).
DeepSeek's cost-efficient V4.1 tier and the first model built on the Causal Encoder-Decoder (CED) architecture: a 552B MoE activating 8B parameters on input and 16B on output.
IBM's compact dense reasoning model for mathematics, code generation, and multilingual dialogue.
Tencent's Hunyuan 4 preview, built for coding agents and complex tool-use workflows.
Z.ai's efficient multimodal model for fast coding and long-horizon agent tasks.
Experimental vision-enabled DeepSeek V4 Flash with image understanding added on top of the 0731 snapshot.
Z.ai's large-scale reasoning flagship for complex software engineering and long-horizon agent tasks.
Open-weight dense 27B vision-language model from Qwen for long-running agent tasks.
The open-weight variant of Qwen3.8 Max, with 95 billion active parameters out of 2.4 trillion total.
The GA release of DeepSeek's open 1.6T-parameter MoE flagship for advanced reasoning and long-horizon agents.
NVIDIA's open 30B MoE built for high-throughput agentic workloads.
Meta's open-weight 30B multimodal model, distilled from Muse Spark for autonomous agents on consumer hardware.
The GA release of DeepSeek's efficiency-optimized V4 Flash, re-post-trained for coding, reasoning, and agent workflows.
Thinking Machines Lab's smaller, more efficient open-weight multimodal model, with 12B active parameters out of 276B total.
Thinking Machines Lab's open-weight multimodal model with 41B active parameters out of 975B total.
MoonshotAI's 2.8T-parameter flagship for complex coding, knowledge work, and long-horizon agent runs.
Z.ai's frontier reasoning model for long-context agents and software engineering.
NVIDIA's open frontier reasoning model for orchestration, coding, and deep research.
MiniMax's multimodal foundation model for long-horizon agentic work.
Open-source 1.6T-parameter MoE with 49B active, built for advanced reasoning and long-horizon agents.
Efficiency-optimized DeepSeek V4 at 284B total / 13B active for fast, high-throughput inference.
MoonshotAI's next-generation multimodal model built for long-horizon coding and multi-agent orchestration.
Efficient Mixture-of-Experts Gemma 4 with only 4B active parameters per token.
Google's fourth-generation open dense model with native multimodal understanding.
Compact 9B vision-language model from the Qwen3.5 family.
Efficient vision-language MoE with linear attention and 3B active parameters.
Dense 27B vision-language model with linear attention for fast response times.
Qwen's largest MoE model with 397B total parameters, 17B active.
Zhipu AI's advanced model with strong reasoning and multimodal support.
NVIDIA's efficient 30B MoE Nemotron model with 3B active parameters.
Open-source model with GPT-5 class reasoning performance.
Efficient 30B MoE vision-language model with 3B active parameters.
Zhipu AI's GLM 4.6 generation model.
Qwen's largest vision-language model with 235B MoE architecture.
Large hybrid reasoning model supporting thinking and non-thinking modes.
OpenAI's budget-friendly reasoning model — fast and surprisingly capable.
OpenAI's compact open-weights model for efficient inference.
Qwen's primary code generation model for software development workflows.
July 2025 update to Qwen3 235B A22B with improved capabilities.
Latest Mistral Small — efficient 24B model for fast, cost-effective tasks.
Updated DeepSeek R1 with performance on par with OpenAI o1.
Efficient 30B MoE language model with 3B active parameters per token.
Mid-size dense Qwen3 model balancing capability and efficiency.
Dense 32.8B parameter model optimized for reasoning and dialogue.
Meta's efficient multimodal MoE model with 16 experts.
Compact open model for efficient on-device or edge deployment.
Mid-size open multimodal model balancing capability and efficiency.
Google's open multimodal model with 128K context and 140+ language support.
January 2025 Mistral Small 24B instruction-tuned model.
Microsoft's compact research model with strong STEM reasoning.
Original DeepSeek V3 — the model that put DeepSeek on the map.
Meta's efficient 70B model matching Llama 3.1 405B on key benchmarks.
Large 72B model from the Qwen 2.5 generation.
NousResearch's 70B model based on Llama 3.1.
NousResearch's large model based on Llama 3.1 405B.
Meta's 70B model from the Llama 3.1 generation.
Compact 8B model from Meta's Llama 3.1 series.
12B model from Mistral and NVIDIA — efficient and multilingual.
