realrouter
Get API Key

Changelog

User-visible product changes only.

pricing

Pricing is now US dollars per million tokens

RealRouter credits are gone. Every model has one published USD rate — exactly one fifth of the model vendor’s own API price — and your balance, the dashboard, and the API all report the same dollars. Plans now state how much usage they include and what rate limit they carry.

  • Balances converted automatically; Basic and Plus subscribers keep the same purchasing power
  • Starter now includes $5.00 of usage, up from the $4.00 its credits were worth
  • Published rate limits per tier: 20, 60, and 200 requests/min with 1, 2, and 4 concurrent requests
  • Breaking: error codes insufficient_credits and request_credit_limit_exceeded are now insufficient_balance and request_cost_limit_exceeded
  • Breaking: API fields ending in _credits are replaced by _usd; /v1/models now returns pricing_usd_per_1m
tools

Computer use and custom tools on /v1/responses

Both tools follow the official OpenAI wire format, so a client written against their docs works here unchanged.

  • computer accepts both {"type":"computer"} and the preview declaration
  • computer_call items carry both action and actions
  • custom supports freeform and Lark-grammar constrained input
  • Vision models only for computer use; send screenshots at 1280px or less on the long edge
models

GPT-5.6 Sol, Terra, and Luna are available

The GPT-5.6 family now spans frontier, balanced, and efficient workloads through the same OpenAI-compatible API.

  • 372K context
  • Sol and Terra support Ultra effort
  • Luna supports up to Max effort
website

Dedicated changelog and docs navigation polish

Changelog now has a permanent page, and the docs sidebar tracks the current section while reading.

  • New /changelog route
  • Docs scroll-spy for tools subsections
  • Header and footer now point to the page
localization

Chinese and English website modes

The website gained a language toggle with typography tuned separately for Chinese copy.

  • Chinese copy across home, docs, pricing, models, dashboard
  • Natural Chinese line breaks
  • Shared dark and light theme support
dashboard

Account console for API keys and credit

Email verification signs users into a private dashboard for key management, balance, billing, and usage.

  • Email code sign-in
  • Self-service API key creation and revoke
  • Usage summary and billing panels
pricing

Subscription plans replace public token buckets

Starter, Basic, and Plus now package monthly credits for API keys and mobile chat.

  • Three public subscription plans
  • Shared credits for API and mobile chat
  • Internal metering remains credit-based
models

gpt-5.5 added to the catalogue

The highest quality tier is available for demanding prompts and important outputs.

  • Text, vision, streaming, tools
  • 256k context tier
  • OpenAI SDK compatible route
platform

Companion mobile app uses the same balance

The same account balance and API key ownership can be used from the companion mobile experience.

  • Same account identity
  • Same balance ledger
  • Designed for quick daily prompts
billing

Cached input billing is live

Repeated prefixes can be billed at the cached input rate when supported by the selected model.

  • Cached input shown separately
  • 1/10x repeated-prefix rate
  • Usage debit tied to successful calls
api

Semantic SSE streaming events

Streaming responses now follow OpenAI-style semantic event shapes for compatible clients.

  • Responses API streaming
  • Chat completion compatibility
  • Tool events where supported