How do I use Requesty with Claude Code?
Point ANTHROPIC_BASE_URL at https://router.requesty.ai with your Requesty key and set ANTHROPIC_MODEL to a Requesty model ID; requests cost provider list price plus Requesty's 5% top-up fee.
# Requesty's documented setup. No /v1: Claude Code adds /v1/messages itself.
export ANTHROPIC_BASE_URL=https://router.requesty.ai
export ANTHROPIC_AUTH_TOKEN=YOUR_REQUESTY_KEY
export ANTHROPIC_API_KEY=""
# Requesty recommends ANTHROPIC_MODEL over the /model picker
export ANTHROPIC_MODEL=anthropic/claude-sonnet-5-5
claude
Shell exports last for one terminal; put the same values in the env block of ~/.claude/settings.json to keep them. Run /status in Claude Code to check the base URL.
Gotchas in Requesty's own docs
| Issue | What happens | What to do |
|---|---|---|
| /model picker | Requesty: "Claude might not accept it sometimes" | Set ANTHROPIC_MODEL instead, or start with --model |
| Two key variables | The Claude Code page uses ANTHROPIC_AUTH_TOKEN, the EU page ANTHROPIC_API_KEY. Claude Code prefers ANTHROPIC_API_KEY when both are set | Set one, and leave ANTHROPIC_API_KEY empty if you use the token |
| Other models | Any Requesty model works with a provider/model ID, for example openai/gpt-6-luna | Expect weaker tool use than Claude on some models; test a real task |
| EU | The global router can run Claude outside the EU | Use https://router.eu.requesty.ai and an EU model, about 10% dearer |
| Empty balance | Requests are rejected with HTTP 402 | Turn on auto top-up or set a key cap you watch |
| Parallel subagents | Requests in flight are capped at five times your balance in dollars below $250 lifetime spend | Keep a balance above a few dollars when subagents fan out |
From docs.requesty.ai integrations and features pages, checked October 13, 2026.
What a request costs through Requesty
A month of typical working days is about $74 on Sonnet 5.5 and $148 on Opus 5.5. A session that sends every step to Opus pays the Opus price for file reads too.
The catch
Anthropic does not support Claude Code through third-party gateways, and its docs warn that a gateway that does not forward new features can break them after an update. Requesty's Claude Code page still uses the older ANTHROPIC_SMALL_FAST_MODEL variable, so check background calls in its logs after setup.
Where Standard Compute fits
Point the same variables at api.stdcmpt.com and keep Claude Code's own model names: smart routing picks a model per request from the ones you allow, sending routine steps to efficient models and hard ones to Claude. One key covers Claude, GPT, Gemini, DeepSeek and GLM, and a flat plan from $19 includes the compute and stops at its budget, with no overage. Teams and agencies get a plan per key and Europe or US rules.
FAQ
What is the Requesty base URL for Claude Code?
https://router.requesty.ai, with no /v1, because Claude Code adds /v1/messages itself. The EU endpoint is https://router.eu.requesty.ai.
Can I use GPT or Gemini in Claude Code through Requesty?
Yes. Set ANTHROPIC_MODEL or --model to a Requesty ID such as openai/gpt-6-luna. Requesty translates the request; how well the model handles Claude Code's tools varies.
Does Requesty work with Codex CLI?
Yes. Requesty documents base_url https://router.requesty.ai/v1 on the Responses API, and says to set model_supports_reasoning_summaries = false because it rejects Codex's reasoning summary field.
Does my Claude Pro or Max plan pay for requests through Requesty?
No. While a gateway key is set, Claude Code sends that key instead of your claude.ai login, and Requesty bills your balance.