Runix Router
Every model, at its public price
The models Runix Router routes to, each at the price its vendor publishes per million tokens. Router's list price is the vendor's public price; contract rates are quoted in writing. Prices as published on September 30, 2026.
02 · Catalogue
The catalogue
11 models from 5 vendors, at standard pay-as-you-go rates: not batch, not off-peak. Where a vendor prices by prompt length, the base tier is shown and the note says where it ends. Every row links to the page the price was read from.
| Model id on Router | Context | Input / 1M | Output / 1M | Cached input / 1M | Source |
|---|---|---|---|---|---|
qwen3-coder-nextQwen3-Coder-Next | 256K | $0.3 | $1.5 | — | Alibaba Cloud pricing · Input up to 32K tokens; 32K to 128K bills $0.50 / $2.50, 128K to 256K $0.80 / $4 (international regions). |
claude-fable-5Claude Fable 5 | 1M | $10 | $50 | $1 | Anthropic pricing |
claude-haiku-4-5Claude Haiku 4.5 | 200K | $1 | $5 | $0.1 | Anthropic pricing |
claude-opus-5Claude Opus 5 | 1M | $5 | $25 | $0.5 | Anthropic pricing |
claude-sonnet-5Claude Sonnet 5 | 1M | $2 | $10 | $0.2 | Anthropic pricing |
minimax-m2.1MiniMax-M2.1 | 200K | $0.3 | $1.2 | $0.03 | MiniMax pricing · On the vendor's legacy list. |
minimax-m2.5MiniMax-M2.5 | 200K | $0.3 | $1.2 | $0.03 | MiniMax pricing · On the vendor's legacy list. |
gpt-5.6-lunaGPT-5.6 Luna | 1.05M | $0.2 | $1.2 | $0.02 | OpenAI pricing · Prompts over 272K input tokens bill at 2× input and 1.5× output. |
gpt-5.6-solGPT-5.6 Sol | 1.05M | $4 | $20 | $0.4 | OpenAI pricing · Promotional price through at least November 21, 2026. Prompts over 272K input tokens bill at 2× input and 1.5× output. |
gpt-5.6-terraGPT-5.6 Terra | 1.05M | $2 | $12 | $0.2 | OpenAI pricing · Prompts over 272K input tokens bill at 2× input and 1.5× output. |
glm-5GLM-5 | 200K | $1 | $3.2 | $0.2 | Z.ai pricing |
USD per million tokens as published by Alibaba Cloud, Anthropic, MiniMax, OpenAI and Z.ai on September 30, 2026. Vendors change prices without notice; the date is the day these were read, and the catalogue is rebuilt from its data file on every deploy. A model whose vendor has withdrawn its public list price is left off until it is re-priced.
03 · Pricing
How a request is priced
Three rules, the same for every model on the list.
List price is the vendor's price
Router's list price for a model is the number on the vendor's own pricing page, input and output separately, per million tokens. There is no separate rate card to reconcile.
auto is priced at the ceiling
A request that lets Router choose the model is billed at the highest list price among the models it may choose from, so choosing auto never costs more than the most expensive explicit choice.
Contract rates in writing
Volume and enterprise rates are quoted per workload and stated in writing before you commit; the catalogue shows list prices only.
04 · Questions
Common questions
Are these Runix prices or the vendors' prices?
Both: Router's list price for each model is the vendor's public price, read from the vendor's own pricing page on the date shown. Contract rates can differ and are quoted in writing.
What does a model id on Router look like?
The vendor's official id where the vendor publishes one, so an OpenAI-compatible client can send it unchanged. The table shows the id as Router accepts it beside the vendor's own name for the model.
Which models can my account use?
The models on this list, subject to the allowed-models setting on your key, which is agreed at intake and can be changed on request.
What about cached input?
Where a vendor publishes a cached-input price, Router bills cached tokens at that price; the column shows it, and a dash means the vendor does not publish one.
Why is a model I expected not on the list?
A model stays on the catalogue only while its vendor publishes a list price for it. When a vendor retires a line, the model comes off until it is re-priced, and keys that named it are told before anything changes.
Put one endpoint in front of all of them
Tell us the models and the volume. A real person replies within one business day with a plan and, where volume warrants it, a rate in writing.
Request access