https://ionrouter.io CumulusCumulus Labs Iondex LeaderboardCategories PlaygroundDocs Get Started Iondex -- LeaderboardIondex -- CategoriesPlaygroundDocs Get Started IonRouter High throughput, low cost inference. Powered by IonAttention. Get StartedPlaygroundEnterprise W26NVIDIA Inception NVIDIA Grace Hopper Superchip IonAttention Engine Not just fast hardware. A faster engine: IonAttention. Our custom inference stack multiplexes models on a single GPU, swaps in ms, and adapts to traffic in real time. Built from the ground up for Grace Hopper. Throughput (tok/s)Single GH200, Qwen2.5-7B IonAttention7,167 Top inference provider~3,000 Read the deep dive Custom Models Bring any model. Get dedicated streams. Deploy your finetunes, custom LoRAs, or any open-source model on our fleet. Dedicated GPU streams with no cold starts and per-second billing. Book a call LoRA Finetune Custom GPU Stream GH2000ms cold start api.ionrouter.io/v1 [?]RoboticsReal-time VLM perception *SurveillanceMulti-stream video analysis #Game GenOn-demand asset generation >AI Video GenText & image to video What Teams Build on Ion From robots to real-time video. Teams use Ion for highest performance robotics perception, multi-camera surveillance, game asset generation, and AI video pipelines. Case Study 5 VLMs, 1 GPU. Five vision-language models on a single GPU -- 2,700 video clips, concurrent users, <1s cold starts. Read the case study API * Zero Code Changes Drop in. Ship faster. Point your existing OpenAI client at Ion. Any language, any framework. One line change. PythonTypeScriptGo Copy from openai import OpenAI client = OpenAI( api_key="sk-your-key-here", base_url="https://api.ionrouter.io/v1" ) response = client.chat.completions.create( model="qwen3.5-122b-a10b", messages=[{ "role": "user", "content": [ {"type": "image_url", "image_url": {"url": "..."}}, {"type": "text", "text": "What's in this image?"} ] }] ) Pricing Models & Pricing Pay per million tokens. No idle costs. GLM-5GLM ZhiPu AI's flagship 600B+ MoE model with state-of-the-art reasoning, coding, and multilingual capabilities, powered by EAGLE speculative decoding on 8x B200 GPUs. ~220 tok/s$1.20 in * $3.50 out Try in Playground glm-5 Kimi-K2.5Kimi MoonShot AI's frontier reasoning model designed for long document understanding, multi-step reasoning chains, and complex problem decomposition across technical and scientific domains. ~120 tok/s$0.20 in * $1.60 out Try in Playground kimi-k2.5 MiniMax-M2.5MiniMax MiniMax's flagship 1M-context language model delivering strong reasoning and instruction following across long documents, multi-turn dialogue, and complex analysis. ~120 tok/s$0.40 in * $1.50 out Try in Playground minimax-m2.5 Qwen3.5-122B-A10BLanguage Cumulus's most capable open-source model -- a 122B MoE with 10B active parameters rivaling leading proprietary models on coding, reasoning, and multilingual benchmarks. ~120 tok/s$0.20 in * $1.60 out Try in Playground qwen3.5-122b-a10b GPT-OSS-120BLanguage A frontier open-source 120B model delivering cutting-edge reasoning and instruction following comparable to leading closed-source systems, ideal for complex agentic workflows and advanced code generation. ~100 tok/s$0.020 in * $0.095 out Try in Playground gpt-oss-120b Wan2.2 Text-to-VideoVideo A 14B text-to-video model optimized for speed via the FastGen runtime, generating clips in under 10 seconds with strong motion coherence. ~8s/clip$0.00194 / GPU*sec Try in Playground wan2.2-t2v-general Flux SchnellImage Black Forest Labs' fastest Flux model, delivering crisp sub-4-second image generation ideal for real-time applications, prototyping, and high-volume pipelines. ~3s/image~$0.005 in * per image out Try in Playground flux-schnell Vision4 more Language9 more Image1 more Video7 more Audio3 more Ready to build? Start in under a minute. No GPU expertise required. Get StartedBook a DemoJoin Discord for $5 Free [cumulus-pixel-logo] Cumulus Labs Privacy Policy*Terms of Service (c) 2026 Cumulus Compute Labs Corporation. All rights reserved.