Skip to content

Solutions / Language models

Your models. Your application.

Build assistants, coding tools, and text workflows with open models running in your own workspace.

  • OpenAI-compatible chat API
  • Choose an open model or bring your own
  • GPU and storage options at launch
Your data stays yours
GPU workspace

Start here

vLLM

Serve a model through an OpenAI-compatible API, with a chat interface for testing.

Launch vLLM

For managed text inference, review the Qwen3.8 27B US endpoint at $1,899/month. This workspace launch is a separate product, with its own hardware and pricing. Managed provisioning is subject to current qualified capacity.

Explore the models

Full library

Qwen

Qwen3.8 27B

Qwen3.8 27B on one dedicated NVIDIA A100 SXM4 80 GB in US-MD-1, United States. $1,899/month per endpoint, subject to current qualified capacity.

Language modelsManaged US endpoint

Qwen

Qwen3 8B

A compact language model for chat, reasoning, and everyday coding tasks.

Language modelsvLLM

Qwen · Unsloth

Qwen3.5-4B

Fine-tune Qwen with Unsloth to route support tickets, detect refund requests, and score urgency.

Language modelsUnsloth

Z.ai

GLM-5.3-Flash

A multimodal model for coding, agents, and document workflows with long context.

Language modelsRequest deployment

Qwen

Qwen3.8-Flash-Next

A sparse multimodal model for coding, visual understanding, and multi-step agent tasks.

Language modelsRequest deployment

Z.ai

GLM-5.3

A flagship language model for complex coding and long-running engineering tasks.

Language modelsRequest deployment

DeepSeek

DeepSeek-V4.1-Flash

Multimodal reasoning and agent workflows with image inputs and million-token context support.

Language modelsRequest deployment

Moonshot AI

Kimi K3

A trillion-parameter multimodal model for long-running coding agents, research, and knowledge work.

Language modelsRequest deployment

MiniMax

MiniMax M3

A multimodal model for coding and agent tasks with text, image, and video inputs.

Language modelsRequest deployment

DeepSeek

DeepSeek-V4-Pro-0813

A flagship DeepSeek model for demanding reasoning, coding, and long-context agent workloads.

Language modelsRequest deployment

Qwen

Qwen3.8-2.4T-A95B

Qwen's open flagship for complex coding, research, and multi-step professional tasks.

Language modelsRequest deployment

Google DeepMind

Gemma 4 26B-A4B

A multimodal mixture-of-experts model for reasoning, coding, and assistants that understand images.

Language modelsvLLM

Cloudflare

Clef-Flash

A multimodal decision model that scores allowed answers for routing, classification, and business workflows.

Language modelsJupyter + PyTorch

See it in action

Document extraction

Explore our invoice demo: structured output, source evidence, and checks in code.

Explore the demo