claude-fable-5-1
| Model | Input | Output | Cache read |
|---|---|---|---|
| List price | $10 | $50 | $0.25 |
| Official line−10% vs list | $9 | $45 | $0.225 |
| Shared pool−50% vs list | $5 | $25 | $0.125 |
TopxAI
Opening page
- Live prices · ai.topxea.com
- PRICING route-usd-v1
- List prices verified 2026-10-02
- SITE v1.0.0
- support@topxea.com
Claude, GPT, Grok, GLM, Kimi and Jev. One base URL.
Claude, GPT, Grok, Kimi and Jev on their official upstreams; GLM as a privately deployed fine-tune. Works with the OpenAI and Anthropic SDKs by changing one address. Flat USD per million tokens, per image or per second of video. Prompts are relayed, never stored.
- 7 vendors
- 14 models
- 7 endpoints
Anthropic3 models
OpenAI4 models
xAI2 models
Zhipu AI1 model
Moonshot AI1 model
TypeSafe1 model
DeepSeek2 models
ai.topxea.com
/v1/messages/v1/chat/completions/v1/responses/v1/images/generations/v1/images/edits/v1/videos/generations/v1/systemoneModels and prices
Two prices per model: the official line is 90% of the vendor's list price, the shared pool 50%. Compare both routes with the list price, in USD per million tokens. Single-route models are priced at their own ratio to the vendor's list price: GLM-5.3-Abliterated (Private deployment) 286%, kimi-k3 (Official line) 80%, jev-1.13.0 (Official line) 107%, deepseek-v4-pro (Official line) 80%, deepseek-flash (Official line) 80%. For models marked with a long-context tier, once the input context reaches the threshold the whole request bills at the higher rate, every token and not just the excess; the threshold and prices are on the line under that model.
Anthropic
FAQ
Terms are the ones published on ai.topxea.com.
Are these the official models?
Claude, GPT, Grok, Kimi and Jev are. ai.topxea.com states that they are served from the official upstream API, with no distilled copies and no third-party mirrors: the official line calls the vendor directly, and the shared pool uses the same upstream at a discounted price. GLM-5.3-Abliterated is the exception: a fine-tuned model on a private deployment that never touches Zhipu's official API. The price table labels it "private deployment" and carries its compliance notice on that row.
What is the difference between the shared pool and the official line?
Same upstream, different routing and price. The official line goes straight to the vendor; the shared pool is discounted: text models cost 50% of the vendor's list price on the shared pool and 90% on the official line. You pick the route when you create a key, or pick "Auto route": one key covers every model, and each request lands on the lowest-priced line that serves its model (shared pool when there is one, otherwise the official line or the model's single route).
How is it billed and how do I top up?
Flat prices in USD per million tokens, input and output counted separately; image models are billed per image in two size tiers (1K and 2K), and the video model per generated second at a rate that depends on the resolution. You prepay a USD balance by card or USDT / USDC. The usage log records model, tokens, cost and latency for every request.
Which clients work with it?
Claude models work with Claude Code and any Anthropic-compatible SDK or client; GPT and Grok models work with Codex CLI and any OpenAI-compatible SDK or client; GLM and Kimi are served on OpenAI-style /v1/chat/completions only; Jev uses TypeSafe's /v1/systemone, reachable from the official TypeSafe SDKs or agent skill by changing one base URL. The gateway exposes OpenAI-style /v1/chat/completions and /v1/responses alongside Anthropic-style /v1/messages.