One-click local proxy that routes Claude Code and Codex to DeepSeek, Qwen, Kimi, GLM, Gemini, Groq — at a fraction of the cost. You bring your API keys. We make it simple.
Real API costs per million tokens. ModelHub adds zero markup — you pay only your provider's rate.
| Provider | Model | Input $/1M | Output $/1M | Savings |
|---|---|---|---|---|
| Anthropic | Claude Opus 4.8 | $5.00 | $25.00 | — |
| Anthropic | Claude Sonnet 4.6 | $3.00 | $15.00 | — |
| OpenAI | GPT-5.5 | $5.00 | $30.00 | — |
| OpenAI | GPT-4o | $10.00 | $30.00 | — |
| Gemini 2.5 Pro | $1.25 | $10.00 | — | |
| ModelHub + DeepSeek | deepseek-v4-flash | $0.14 | $0.28 | Save 99% |
| ModelHub + DeepSeek | deepseek-v4-pro | $0.44 | $0.87 | Save 98% |
| ModelHub + Kimi | kimi-k2.5 | $0.60 | $3.00 | Save 88% |
| ModelHub + GLM | glm-4-plus | $0.50 | $0.80 | Save 97% |
| ModelHub + Qwen | qwen-max | $0.80 | $1.20 | Save 96% |
| ModelHub + Gemini | gemini-2.5-flash | $0.15 | $0.60 | Save 98% |
| ModelHub + Groq | groq-llama-70b | $0.59 | $0.79 | Save 97% |
| ModelHub + SiliconFlow | sf-deepseek-v3 | $0.27 | $0.55 | Save 98% |
Real problems developers search for — solved by ModelHub.
Anthropic API costs add up fast. Route Claude Code to DeepSeek and cut your bill by 99%. Same tool use, same workflow.
Set up a local proxy for Claude Code in 2 minutes. No Python, no Docker. Just download the EXE and run.
Region restrictions or billing issues? ModelHub runs locally with your own API keys. No more account problems.
Switch between DeepSeek, Qwen, Kimi, GLM, Gemini, Groq without restarting. One panel to rule them all.
Your code and API keys never leave your machine. Privacy-first architecture. Data doesn't cross third-party servers.
Codex users: route to DeepSeek, Qwen, or GLM via ModelHub. Full Responses API + streaming included.
Download (Mac or Windows) and double-click. No Python, no npm, no Docker. It just works.
DeepSeek, Qwen, GLM, Kimi, Gemini, Groq, SiliconFlow, OpenRouter.
Everything runs on your machine. API keys never leave. Zero data leakage.
Switch between 24 models without restarting Claude Code or Codex.
File read/write, command execution — all tool calls work perfectly.
Full management panel at localhost:4000. Manage keys, switch models, view logs.
No hidden fees. No token markups. No subscriptions required.
Available for Mac & Windows. Double-click to run. No dependencies. Works everywhere.
Sign up at DeepSeek, SiliconFlow, or any supported provider. Paste your key into the web UI.
Claude Code or Codex connects automatically. Using DeepSeek for 97% less. Zero config.
Refer a friend. When they buy, you get free time. No limit — stack up.
No. ModelHub runs entirely on your local machine as a desktop EXE. Nothing leaves your computer.
DeepSeek, SiliconFlow, OpenRouter, Zhipu GLM, Kimi (Moonshot), Google Gemini, Alibaba Qwen, and Groq.
Yes. Full Claude Code tool use — file read/write, command execution, sub-agents — works perfectly.
ModelHub fully supports Codex via the Responses API. Same setup, same savings.
Yes. Download the free 7-day trial. No credit card required. Full features with DeepSeek enabled.