Docs / Changelog
Changelog
Notable updates to RelayRouter, newest first.
#2026-10
Models
- Claude:
claude-opus-5-5,claude-opus-5,claude-sonnet-5-5andclaude-fable-5-1are in the catalog - GPT: the GPT-6 family (
gpt-6,gpt-6-astra,gpt-6-luna,gpt-6-sol,gpt-6.1-sol) and GPT-5.6 (gpt-5.6-sol,gpt-5.6-luna,gpt-5.6-terra) - Gemini:
gemini-3.6-flash,gemini-3.7-flashandgemini-3.8-flash - Kimi K3:
kimi-k3is available on the Anthropic protocol (POST /v1/messages) - Retired from the catalog: older dated snapshots, for example the GPT-4 and GPT-4o dated releases, the o1 and o3 dated releases, and Claude 3.5 and 3.7
Pricing
- GPT group settlement rate is now ¥0.6 per $1 of standard usage, about 91% below the ¥6.8/$ market reference
- Kimi group now settles at ¥1.75 per $1:
kimi-k2.6is ¥5.43 per 1M input tokens and ¥22.85 per 1M output tokens - DeepSeek now bills at 1.1x DeepSeek's official time-of-day prices, following its new peak and off-peak pricing. Off-peak,
deepseek-v4-flashis ¥1.1 per 1M input tokens, ¥4.4 per 1M output tokens and ¥0.022 per 1M cache-hit tokens, anddeepseek-v4-prois ¥4.95, ¥14.85 and ¥0.165. The whole request costs twice that on weekdays 09:00 to 12:00 and 14:00 to 18:00 Beijing time - The models page (
/models) andllms.txtare regenerated from the live gateway catalog and rates
#2026-07
Pricing
- The models page lists every public group with its settlement rate and its discount against the ¥6.8/$ market reference, plus per-1M-token prices for the direct groups (DeepSeek, Kimi, MiniMax, Z.ai)
Site
- Chinese version of the site at
/zh
#2026-06
Protocols
- Anthropic compatible
POST /v1/messages(set the Anthropic SDK base_url to the site root) and Gemini compatiblePOST /v1beta/models/{model}:generateContent, alongside OpenAI chat completions
Documentation
- Docs refreshed for accuracy: every endpoint, parameter and model ID in these pages now matches the live API
- Per-site documentation: each brand documents exactly the endpoints it serves
#Initial release
Platform
- OpenAI-compatible chat completions endpoint covering models from OpenAI, Anthropic, Google DeepMind, DeepSeek, MiniMax and Moonshot AI
- Image generation via the OpenAI-compatible images endpoint
- Single API key and single base URL across all providers
- Console with API key management and usage tracking