# EcoHash

> EcoHash provides RTX Pro 6000 GPU cloud, dedicated inference endpoints, and
> OpenAI-compatible model APIs for open-model inference, fine-tuning, voice,
> image, RAG, and coding workloads. Base URL: https://ecohash.com. API: https://api.ecohash.com/v1.

## RTX Pro 6000 GPU cloud
- [RTX Pro 6000 GPU rental](https://ecohash.com/gpu/rtx-pro-6000): 96GB Blackwell GPU workspaces and dedicated endpoints for open-model inference, fine-tuning, rendering, and batch workloads. 1/2/4/8-GPU configurations, 16 vCPU and 64GB RAM per GPU, billed per second.
- [Pricing](https://ecohash.com/pricing): current GPU, model API, and storage rates, pulled live from the billing database.
- [GPU availability feed](https://api.ecohash.com/platform/gpu-availability): public JSON, no authentication. Live per-configuration specs, price and stock status — the authoritative source for "is it available now", ahead of any third-party listing.
- [GPU pricing feed](https://api.ecohash.com/platform/gpu-prices): public JSON, no authentication. Current per-GPU-hour rates.

## Products

### AI Infrastructure
- [GPU Instances & Clusters](https://ecohash.com/gpu/rtx-pro-6000): RTX Pro 6000 96 GB instances with 1/2/4/8 GPUs (SSH, JupyterLab, browser terminal, auto-expiry) and clusters of 2–8 identical replicas behind one load-balanced endpoint. Billed per second.
- [Storage](https://ecohash.com/storage): Cloud Drives (block storage, one instance at a time) and Shared Filesystems (CephFS, mounted by many instances or every cluster replica). 50–500 GB, survives instance termination, export to a download link.

### Datasets
- [Datasets](https://ecohash.com/datasets): reusable, validated JSONL training datasets — chat-format rows or seed prompts synthesized with an open teacher model.

### Training
- [Fine-Tuning](https://ecohash.com/fine-tuning): LoRA / QLoRA adapters on 1–4 GPUs, platform or bring-your-own base model, config templates, per-minute billing.

### Inference
- [Dedicated Inference](https://ecohash.com/dedicated-inference): your models on reserved GPU capacity, same OpenAI-compatible API, billed per GPU-hour.
- [Inference API](https://ecohash.com/inference): OpenAI-compatible API across text, vision, embeddings, image, speech and video, billed per token.
- [Models](https://ecohash.com/models): the live model catalog.

## Use cases
- [Voice & Speech](https://ecohash.com/use-cases/voice-speech): STT and TTS models.
- [Image Generation](https://ecohash.com/use-cases/image-generation): open image models.
- [RAG & Search](https://ecohash.com/use-cases/rag): embeddings and rerankers.
- [Conversational AI](https://ecohash.com/use-cases/conversational-ai): open chat models.

## Solutions
- [Solutions overview](https://ecohash.com/solutions): end-user products built on EcoHash infrastructure.
- [AI Video Studio](https://ecohash.com/solutions/videos): EcoHash Studio at videos.ecohash.com — generative video, avatars, voice cloning and a timeline editor, billed by the second.
- [AI Coding](https://ecohash.com/solutions/coding): EcoHash Coding at coding.ecohash.com — flat monthly plans for coding tools on GLM, DeepSeek, MiniMax and Kimi.
- [Digital Employees](https://ecohash.com/solutions/digital-employees): Octok at octok.com — AI employees for global expansion, built on EcoHash inference.

## Models
- [Llama-3.1-8B-Instruct](https://ecohash.com/models/llama-3.1-8b-instruct): `llama-3.1-8b-instruct`
- [Kokoro-82M](https://ecohash.com/models/kokoro-82m): `kokoro-82m`
- [Qwen3-ASR-1.7B](https://ecohash.com/models/qwen3-asr-1-7b): `qwen3-asr-1-7b`
- [Gemma-4-31B-IT](https://ecohash.com/models/gemma-4-31b-it): `gemma-4-31b-it`
- [Qwen3-TTS](https://ecohash.com/models/qwen3-tts): `qwen3-tts`
- [Jina-Embeddings-V3](https://ecohash.com/models/jina-embeddings-v3): `jina-embeddings-v3`
- [Jina-Embeddings-V4](https://ecohash.com/models/jina-embeddings-v4): `jina-embeddings-v4`
- [BGE-Reranker-V2-M3](https://ecohash.com/models/bge-reranker-v2-m3): `bge-reranker-v2-m3`
- [Z-Image-Turbo](https://ecohash.com/models/z-image-turbo): `z-image-turbo`
- [Seedance 2.0](https://ecohash.com/models/ecolink-video-gen-2.0): `ecolink-video-gen-2.0`
- [DeepSeek-V4-Flash](https://ecohash.com/models/DeepSeek-V4-Flash): `DeepSeek-V4-Flash`
- [DeepSeek-V4-Pro](https://ecohash.com/models/DeepSeek-V4-Pro): `DeepSeek-V4-Pro`
- [qwen3-embedding-0.6b](https://ecohash.com/models/qwen3-embedding-0.6b): `qwen3-embedding-0.6b`
- [qwen3-omni-30b-a3b-instruct](https://ecohash.com/models/qwen3-omni-30b-a3b-instruct): `qwen3-omni-30b-a3b-instruct`
- [Fun-ASR-Nano](https://ecohash.com/models/fun-asr-nano): `fun-asr-nano`
- [GLM-5.2](https://ecohash.com/models/GLM-5.2): `GLM-5.2`
- [qwen3-vl-8b-instruct](https://ecohash.com/models/qwen3-vl-8b-instruct): `qwen3-vl-8b-instruct`
- [qwen3-coder-30b-a3b-instruct](https://ecohash.com/models/qwen3-coder-30b-a3b-instruct): `qwen3-coder-30b-a3b-instruct`
- [gpt-oss-20b](https://ecohash.com/models/gpt-oss-20b): `gpt-oss-20b`
- [ViiTorVoice-NAR](https://ecohash.com/models/viitor-voice-nar): `viitor-voice-nar`
- [Whisper-Large-V3-Turbo](https://ecohash.com/models/whisper-large-v3-turbo): `whisper-large-v3-turbo`
- [FLUX.2 Klein](https://ecohash.com/models/flux2-klein): `flux2-klein`
- [Qwen-Image](https://ecohash.com/models/qwen-image): `qwen-image`
- [Qwen3.6-27B](https://ecohash.com/models/qwen3.6-27b): `qwen3.6-27b`
- [Qwen3.6-35B-A3B](https://ecohash.com/models/qwen3.6-35b-a3b): `qwen3.6-35b-a3b`
- [Wan2.2-T2V-A14B](https://ecohash.com/models/wan22-t2v-a14b): `wan22-t2v-a14b`
- [Wan2.2-S2V-14B (Presenter / Talking Head)](https://ecohash.com/models/wan22-s2v-14b): `wan22-s2v-14b`
- [Stable Audio Open 1.0](https://ecohash.com/models/stable-audio-open): `stable-audio-open`
- [pyannote Speaker Diarization 3.1](https://ecohash.com/models/pyannote-diarization-3-1): `pyannote-diarization-3-1`
- [LTX 2.5](https://ecohash.com/models/ltx-2-5): `ltx-2-5`
- [Qwen3.8-27B](https://ecohash.com/models/qwen3.8-27b): `qwen3.8-27b`
- [Kimi-K3](https://ecohash.com/models/Kimi-K3): `Kimi-K3`
- [GLM-5.3](https://ecohash.com/models/GLM-5.3): `GLM-5.3`
- [MiniMax-M3](https://ecohash.com/models/MiniMax-M3): `MiniMax-M3`
- [qwen3.8-max](https://ecohash.com/models/qwen3.8-max): `qwen3.8-max`
- [Wan 3.0](https://ecohash.com/models/wan3.0-video): `wan3.0-video`
- [wan3.0-video-prime](https://ecohash.com/models/wan3.0-video-prime): `wan3.0-video-prime`
- [Seedance 2.5](https://ecohash.com/models/dreamina-seedance-2-5-260628): `dreamina-seedance-2-5-260628`
- [qwen3.8-flash](https://ecohash.com/models/qwen3.8-flash): `qwen3.8-flash`

## Blog
- [Blog index](https://ecohash.com/blog): tutorials, model-fit notes, benchmark summaries, and product updates.
- [RSS feed](https://ecohash.com/blog/rss.xml)
- [How to deploy and tune Qwen3.8-27B on one RTX Pro 6000](https://ecohash.com/blog/deploy-qwen38-27b-rtx-pro-6000): Qwen3.8-27B is 27B dense in FP8, so one RTX Pro 6000 holds it with room for a large KV
cache. The serving config EcoHash runs in production, the five settings whose obvious
alternative fails, and the one flag that doubled what a single card serves.
- [Run your Retell AI agent on an open model: the custom LLM bridge](https://ecohash.com/blog/run-your-retell-ai-agent-on-an-open-model-the-custom-llm-bridge): How to connect a Retell AI voice agent to EcoHash's OpenAI-compatible API with a small WebSocket bridge, what it does to the per-minute LLM bill, and what stays on Retell.
- [Give your Vapi voice agent a cheaper voice: Kokoro TTS via custom-voice](https://ecohash.com/blog/vapi-custom-tts-kokoro): How to plug EcoHash's Kokoro TTS into a Vapi voice agent as a custom voice: a small adapter server, the assistant config, a local test, and what it costs per minute compared to ElevenLabs.
- [Build a voice agent with Whisper, Kokoro, and an OpenAI-compatible API](https://ecohash.com/blog/kokoro-whisper-voice-agent): How to build a speech-to-text to LLM to text-to-speech voice agent on EcoHash using Whisper, a chat model, and Kokoro, all through one OpenAI-compatible API key.
- [Qwen3 Coder 30B on RTX Pro 6000: API, dedicated endpoint, or GPU workspace?](https://ecohash.com/blog/qwen3-coder-30b-rtx-pro-6000): Three ways to run Qwen3 Coder 30B on EcoHash — the shared OpenAI-compatible API, a dedicated endpoint, or an RTX Pro 6000 GPU workspace — and how to choose between them.
- [Best models to run on RTX Pro 6000 96GB](https://ecohash.com/blog/best-models-for-rtx-pro-6000): A task-by-task guide to choosing open models for a 96GB RTX Pro 6000 on EcoHash: coding, general chat, embeddings and rerankers for RAG, and speech models, all through an OpenAI-compatible API.
- [What makes RTX Pro 6000 96GB a good GPU for AI inference?](https://ecohash.com/blog/rtx-pro-6000-gpu-advantages): How the NVIDIA RTX Pro 6000 Blackwell Server Edition with 96GB VRAM fits open-model inference on EcoHash, which model sizes it suits, how it compares to other GPUs, and how to access it.

## Company
- [About](https://ecohash.com/about): EcoHash Platform Technology LLC builds distributed AI inference from energy assets. A subsidiary of Cango Inc. (NYSE: CANG), founded in 2025.
- [Contact](https://ecohash.com/contact): Contact EcoHash sales about the inference API, on-demand GPU instances and dedicated model endpoints: sales@ecohash.com.

## Legal and compliance
- [Service Level Agreement](https://ecohash.com/legal/sla): 99% monthly reliability and committed throughput on Dedicated Inference Instance Endpoints, plus P0-P3 support response targets.
- [Data Processing Agreement](https://ecohash.com/legal/dpa): Covers EcoHash's processing of personal data on behalf of customers under GDPR, UK GDPR and similar privacy laws.
- [Privacy Policy](https://ecohash.com/privacy): How EcoHash collects, uses and protects personal data across the website, console, inference APIs and GPU services.
- [Terms of Service](https://ecohash.com/terms): Terms governing use of the website, console, inference APIs, dedicated model endpoints and GPU instance or cluster services.
- [Cookie Policy](https://ecohash.com/cookies): Which cookie categories the website uses and how to control them.

## Full reference
- [llms-full.txt](https://ecohash.com/llms-full.txt): every model with pricing, plus blog posts with summaries.