for overseas ai teams · 2026

One key for
every coding agent.

Claude Code, OpenClaw, Codex CLI, and Gemini CLI all work against obixhub unchanged. Point the base URL at obixhub, drop in your obx_key, and put Claude, GPT, and Gemini on one prepaid balance. We don't keep your prompts.

start with $50 →read the docs

request topology4 clients · one base url

claude code·openclaw   codex cli     gemini cli
        │                  │              │
  /v1/messages     /chat/completions  /v1beta/…:gen
        └──────────────────┼──────────────┘
                           ▼
              ┌────────────────────────┐
              │      obixhub.com       │
              │  one base url · usd    │
              └───────────┬────────────┘
                          │
              ┌───────────┼───────────┐
              ▼           ▼           ▼
          ┌───────┐   ┌───────┐   ┌───────┐
          │  ant  │   │  oai  │   │  ggl  │
          │claude │   │  gpt  │   │gemini │
          └───────┘   └───────┘   └───────┘

point your cli here

Four CLIs. Two env vars each.

Each of the four works as-is — no plugin to install, no fork to maintain. Set the base URL, drop in your obx_ key, keep using the CLI you already use.

claude codespeaks anthropic messages

POST /v1/messages

# point claude code at obixhub
export ANTHROPIC_BASE_URL="https://obixhub.com"
export ANTHROPIC_API_KEY="obx_..."

$ claude

primary upstreamanthropic/★

openclawspeaks anthropic messages

POST /v1/messages

# point openclaw at obixhub
export ANTHROPIC_BASE_URL="https://obixhub.com"
export ANTHROPIC_API_KEY="obx_..."

$ openclaw

primary upstreamanthropic/★

codex clispeaks openai chat completions

POST /chat/completions

# point codex cli at obixhub
export OPENAI_BASE_URL="https://obixhub.com"
export OPENAI_API_KEY="obx_..."

$ codex

primary upstreamopenai/★

gemini clispeaks google gemini api

POST /v1beta/…:generateContent

# point gemini cli at obixhub
export GEMINI_BASE_URL="https://obixhub.com"
export GEMINI_API_KEY="obx_..."

$ gemini

primary upstreamgoogle-ai-studio/★

anything else that speaks these wire formats works tooanthropic sdkopenai sdkvercel ai sdkcursoraidercline

models reachable from any of the four clientsfull catalog →

anthropic/claude-opus-4-7

anthropic/claude-sonnet-4-6

openai/gpt-5

openai/gpt-4o

google-ai-studio/gemini-2.5-pro

google-ai-studio/gemini-2.5-flash

pricing

Pay-as-you-go. Prepaid. No surprises.

You're billed for the exact tokens each model reports, at a flat rate just over the provider's published price — the same simple basis whether you're on Claude, GPT, or Gemini. Top up once, spend down, top up again.

minimum top-up$50.00 USD
free creditsnone
paymentstripe (usd)
subscriptionno
refundable balanceyes · on request

what you can count on

→Exact provider token counts — no inflation, no rounding up.
→One flat rate over list, identical across all three model families.
→Prepaid, so a bill can never surprise you.
→Cache savings passed straight through to your balance.

privacy posture

Your prompts and completions never land in our storage by default.

We record only metadata — token counts, latency, model, and status. Never message bodies, never tool inputs. Your conversations pass through and are gone.

Teams that need detailed logs for SOC 2 or customer due diligence can opt in per API key, kept in private storage you control — and even then, never shared with the model providers.

what you get

one balanceClaude, GPT, and Gemini usage on a single prepaid balance and one invoice.

swap by nameMove between models and providers by changing one string — no code rewrite.

caching passed throughWhen a model serves cached tokens, the savings land on your balance, not ours.

usage you can seePer-model spend, latency, and token counts in your dashboard.

Top up $50. Get a key. Ship today.

≈ 60 seconds from signup → first stream

create account →

One key forevery coding agent.

Four CLIs. Two env vars each.

Pay-as-you-go. Prepaid. No surprises.

Top up $50. Get a key. Ship today.

One key for
every coding agent.