reAPI: connect your own account

Set up

Set up reAPI in Claude Code, Codex or any agent

1Give this to your agent

in your agent's chat

set up treg — https://treg.sciatafrica.com/llms.txt

2Or add the MCP server yourself

claude code

claude mcp add --transport http treg https://treg.sciatafrica.com/mcp

ChatGPT

Settings → Connectors → add https://treg.sciatafrica.com/mcp as a custom connector.

Grok bot

Add https://treg.sciatafrica.com/mcp as a remote MCP connector in its tool settings.

Claude Desktop

Settings → Connectors → Add custom connector → https://treg.sciatafrica.com/mcp.

Cursor, Codex, any MCP client

Point it at https://treg.sciatafrica.com/mcp (HTTP transport).

CLI

curl -fsSL https://treg.sciatafrica.com/install.sh | sh

Plain HTTP

LangChain, CrewAI or any code: /call/<tool-id> + X-Treg-Token: <token>. No SDK.

Why treg.to

Why call reAPI through treg.to

Your account, held safely

Connect once; treg.to keeps the credential server-side and injects it per call. No key on any machine.

Price before the call

The provider's own rate, $0.000 markup. How billing works.

Account billing

Calls through your own connection are not metered by treg.

Your account connection

Connect your account with the provider's supported authentication method.

Switch by changing a word

Another provider is a different word in the prompt, not a new integration.

One key, the whole catalog

The same token calls Google AI (Gemini API), OpenRouter, PiAPI and 3,611 other tools.

The shelf

All 6 reAPI tools

Image generation · 3 tools

Google: Gemini 3 Pro Image (Nano Banana Pro) · Own connection required · live-verified · $0.03-$0.034/success
Generate or edit one image with Gemini 3 Pro Image at 1K, 2K or 4K, from a prompt and up to 14 reference images. reapi.image-gen.gemini-3-pro-image
Default channel; live 2026-09-14 one 1K image settled 30 credits (4K documented only). The -official Google channel ($0.108 / $0.192) is token-metered and not listed. One image per request.
OpenAI: GPT Image 2 · Own connection required · live-verified · $0.03-$0.081/success
Generate or edit one image with GPT Image 2 at a flat per-resolution price. reapi.image-gen.gpt-image-2
One image per request (n is fixed at 1). Failed or moderated requests are not charged. Live 2026-09-14: one 1k image settled 30 credits; 2k and 4k are documented only.
OpenAI: GPT Image 2.5 (Flare / Sunburst) · Own connection required · live-verified · $0.004-$0.41/success
Generate or edit images with GPT Image 2.5. Flare is the fast default; Sunburst is the precise editor. The -official models are OpenAI's own metered channel, priced by quality and size; the flat-rate channel had no upstream capacity when listed. reapi.image-gen.gpt-image-2-5
Official channel cells are the credits observed live on 2026-09-14 (5/4/4 low, 10/8/8 medium, 40/31/31 high for 1024x1024 / 1024x1536 / 1536x1024; Flare and Sunburst meter identically). Upstream is token-metered, so a very long prompt adds a fraction of a cent the table does not carry. Flat-rate channel rows ($0.0173 Flare / $0.023 Sunburst per image) are documented only: both answered 'No available channel' upstream at listing time and charged nothing.

Video generation · 3 tools

Check a reAPI task · Own connection required · helper · live-verified · free
Read a generation task's status, output URLs and the credits it consumed (1 credit = $0.001). reapi.tasks.get
Polling does not consume credits.
ByteDance: Seedance 2.5 · Own connection required · live-verified · $0.474-$13.9/success
Create a Seedance 2.5 video task from a prompt, optionally guided by up to 30 reference images and 10 audio tracks, with synced generated audio. reapi.video-gen.seedance-2-5
Billed per second of output; failed or moderated tasks are refunded in full. 480p and 720p rates are the credits observed live on 2026-09-14 (593 credits for 5 s at 480p; 4,803 for 18 s at 720p), which sit just under the pricing page's rounded $0.119 / $0.267. Video references (video_urls) add their own length to the billable seconds and are deliberately not listed so the table stays exact.
ByteDance: Seedance 2.5 (Less Restriction) · Own connection required · live-verified · $0.474-$13.9/success
Less Restriction: the same Seedance 2.5 model with content_filter off, so a photo of a real person is accepted as the subject reference and a voice clip as the speech reference; the default row refuses real faces. Same route and price; illegal content is still refused. reapi.video-gen.seedance-2-5.unrestricted
Same meter as the default row: per second of output, refunded on failure. Live 2026-09-14: an 18 s 720p clip from one face image plus one voice clip settled 4,803 credits ($4.80); the 5 s 480p probe from a reference image settled 593.

Reliability badges come from live traffic through treg.to. Live-verified means a successful provider check; it does not verify every request variant.

Questions

Before you start

Do I need a reAPI account?

Yes. This is an own-account connection: you sign in to reAPI once and your agent uses that connection through your treg.to token. Calls on it are never metered.

How do I add reAPI to Claude Code?

Run: claude mcp add --transport http treg https://treg.sciatafrica.com/mcp. One MCP server carries reAPI and the rest of the catalog.

Which frameworks does it work with?

Anything that speaks MCP (Claude Code, Claude Desktop, ChatGPT, Codex, Cursor, Grok) and anything that can make an HTTP request (LangChain, CrewAI, LlamaIndex, plain code).

Is this the official reAPI MCP server?

No. treg.to serves reAPI's real API through its own metered proxy: the request is the provider's own, the credential is injected server-side, and the answer is relayed verbatim. The official reAPI channels are linked above.