Detailed Category Table
API Group Type Comparison Guide
Group Types and Features
| Group | Type | Rate | Supported Models |
|---|---|---|---|
| default | Mixed ChatGPT (Azure channel) + Claude (reverse channel) + MJ (fast models) + Chinese models (DeepSeek+Qwen) | official rate ×1 | openAI, Claude, Chinese models |
| Enterprise high-availability LLMs | Chinese models (DeepSeek+Qwen) | official rate ×1 | Chinese models |
| Premium grok | grok models (subset of grok models) | official rate ×5 | grok |
| Premium gemini | Gemini (Google channel) models | official rate ×1 | Gemini |
| Official relay gemini | Gemini (Google channel) models, higher price (more accounts, higher cost) | official rate ×3 | Gemini |
| Azure only | ChatGPT models (Azure channel) + Chinese models (Doubao+deepseek) only | official rate ×1.5 | openAI, Claude, Chinese models |
| Official relay | ChatGPT (Azure channel) + ChatGPT (official relay channel) + Chinese models (Azure first, official relay when unavailable) | official rate ×3 | openAI, Chinese models |
| Official relay OPENAI | ChatGPT models (official relay channel) + Azure channel (fallback) | official rate ×6 | openAI |
| Premium official relay OPENAI | ChatGPT models (official relay channel), higher price (more accounts) | official rate ×8 | openAI |
| Reverse | Supports GPT + Claude + Gemini + Grok | official rate ×1.4 | openAI, Claude |
| Limited-time special | Chinese models (DeepSeek+Qwen) + Gemini (Google channel) + ChatGPT (Azure channel) | official rate ×0.6 | Gemini, Chinese models |
| Official relay Claude 2 | Claude (AWS official relay channel) | official rate ×6 | Claude |
| Official relay Claude 3 | Claude (AWS official relay + Anthropic official relay channels) | official rate ×12 | Claude |
| Direct Claude | Claude (Anthropic official relay channel) | official rate ×16 | Claude |
| Claude code exclusive | Claude code | official rate ×1.5 | Claude code |
Pros and Cons Comparison
| Feature | Reverse | Default (mixed) | Azure | Official relay | Official relay / Premium official relay OPENAI | Premium gemini | Official relay gemini | Official relay Claude | Direct Claude |
|---|---|---|---|---|---|---|---|---|---|
| Pros | Same as ChatGPT website, smarter, lower price; replies differ somewhat from the API—even for the same model, likely different versions from the provider | Same as ChatGPT website, smarter, lower price | Many channels, strong redundancy, high volume and concurrency, no moderation, supports FC and TC | Fast, high volume and concurrency, no moderation, supports FC and TC | Same as ChatGPT website, fast, high volume and concurrency, more accounts, more stable | Same as official site, smarter, lower price, high volume, stable | Same as official site, smarter, lower price, high volume, more stable, more accounts | From Amazon and Anthropic official sources; many channels, strong redundancy, high volume and concurrency | From Anthropic official sources; more accounts, more stable |
| Cons | Non-official API | Non-official API | Azure may lag at times | Affected by official API stability | Affected by official API stability | Affected by official API stability | Higher cost, higher multiplier, affected by official API stability | Affected by official API stability | Affected by official API stability |
| Overview | Uses OpenAI official reverse; avoid frequent health checks such as sayl | Mix of reverse, Azure, and official relay | Enterprise Azure without moderation; supports FC and TC, 10M TPM; OpenAI used as fallback on failure | OpenAI official channels; supports FC and TC, 10M TPM; GPT-3.5 uses high-limit OpenAI accounts, high concurrency | OpenAI official channels; supports FC and TC, 10M TPM; GPT-3.5 uses high-limit OpenAI accounts, high concurrency | Google official API, high-limit accounts, high concurrency | Google official API, high-limit accounts, high concurrency | Official channels; supports function calling and Claude code; high-limit accounts, high concurrency | Official channels; supports function calling and Claude code; high-limit accounts, high concurrency |
Model Support
| Model Type | Reverse | Default (mixed) | Azure | Official relay |
|---|---|---|---|---|
| GPT-4 | Reverse, full support | Reverse, full support | Azure, full support | OpenAI, full support |
| GPT-3.5 | Reverse, full support | Reverse, full support | Azure, full support | OpenAI, full support |
| Other OpenAI base models | Full support | Full support | Full support | Full support |
| Midjourney | Not supported | Full support | Not supported | Not supported |
| Chinese models | Not supported | Full support | Partial support | Not supported |
| Claude | Reverse, full support | Full support | Full support | Reverse, full support |
Channel Descriptions
- ChatGPT official relay channel: from openai.com
- Azure channel: from Microsoft Azure
- Gemini Google channel: from Google
- Chinese models: from each provider’s official source
- Claude AWS official relay: from Amazon
- Claude Anthropic official relay: from Anthropic official
You can set the group on the token page via token settings.