Model Catalog
AIOrouter is an All-In-One AI Router — providing a single OpenAI-compatible API endpoint that routes to Western models (Google Gemini, Anthropic Claude), Chinese models (DeepSeek, Alibaba Qwen, Moonshot Kimi, Zhipu GLM), and more — all with built-in privacy protection (bidirectional PII pseudonymization, technical secret redaction, AI Firewall) and Canada-resident infrastructure.
LLM Public Token Pricing
Prices below are public provider rates in USD per 1 million tokens before CAD conversion, taxes, and any account-specific retail presentation. The Dashboard and API model response are the customer-facing source for your exact billable rate.
| Model | Input (USD/1M tokens) | Output (USD/1M tokens) | Cache pricing |
|---|---|---|---|
| claude-fable-5 (suspended) | $10.00 | $50.00 | Cache write: $12.50; Cache read: $1.00 |
| claude-haiku-4.5 | $1.00 | $5.00 | Cache write: $1.25; Cache read: $0.100 |
| claude-opus-4.7 | $5.00 | $25.00 | Cache write: $6.25; Cache read: $0.500 |
| claude-opus-4.8 | $5.00 | $25.00 | Cache write: $6.25; Cache read: $0.500 |
| claude-sonnet-4.6 | $3.00 | $15.00 | Cache write: $3.75; Cache read: $0.300 |
| deepseek-v4-flash | $0.140 | $0.280 | Implicit cache: $0.0028 |
| deepseek-v4-flash-alibaba | $0.138 | $0.275 | Cache read: $0.028 |
| deepseek-v4-pro | $0.435 | $0.870 | Implicit cache: $0.0036 |
| gemini-2.5-flash | $0.300 | $2.50 | Implicit cache: $0.030 |
| gemini-2.5-pro | $1.25 up to 200K / $2.50 above | $10.00 | Implicit cache: $0.125 |
| gemini-embedding-2 | $0.025 | $0.0000 | Not configured |
| glm-5 | $0.180 | $0.720 | Not configured |
| glm-5.1 | $0.825 | $3.30 | Cache read: $0.083; Implicit cache: $0.165; explicit creation $1.03 |
| glm-5.2 | $1.40 | $4.40 | Implicit cache: $0.260 |
| kimi-k2.6 | $0.950 | $4.00 | Implicit cache: $0.160 |
| kimi-k2.7-code | $0.950 | $4.00 | Implicit cache: $0.190 |
| qwen-deep-research | $7.74 | $23.37 | Not configured |
| qwen-flash-character | $0.034 | $0.203 | Cache read: $0.0070 |
| qwen3-235b | $0.700 | $2.80 non-thinking / $8.40 thinking | Not configured |
| qwen3.5-omni-flash-realtime | $0.110 | $0.440 | Not configured |
| qwen3.5-omni-plus-realtime | $0.550 | $2.20 | Not configured |
| qwen3.5-plus-2026-04-20 | $0.573 | $0.688 | Cache read: $0.011; explicit creation $0.359 |
| qwen3.6-27b | $0.413 | $2.48 | Not configured |
| qwen3.6-35b-a3b | $0.248 | $1.49 | Not configured |
| qwen3.6-flash | $0.165 | $0.990 | Cache read: $0.017; Implicit cache: $0.033; explicit creation $0.206 |
| qwen3.6-flash-2026-04-16 | $0.165 | $0.990 | Cache read: $0.017; explicit creation $0.206 |
| qwen3.6-plus | $0.276 | $1.65 | Cache read: $0.028; Implicit cache: $0.055; explicit creation $0.345 |
| qwen3.6-plus-2026-04-02 | $0.276 | $1.65 | Cache read: $0.028; explicit creation $0.345 |
| qwen3.7-max | $1.65 | $4.95 | Cache read: $0.165; Implicit cache: $0.330; explicit creation $2.06 |
Cache-Aware Billing
When a provider reports cached input tokens, AIOrouter uses the provider's public cache price instead of treating those tokens as regular input. The same calculation applies to subscription allowance, Top-Up balance, and prepaid compatibility billing paths.
- Alibaba Qwen and Alibaba GLM may report implicit cache hits, explicit cache reads, and explicit cache-creation tokens. Explicit cache creation can cost more than regular input on providers that price creation separately.
- Gemini implicit cache hits are passed through when Gemini reports cached input token counts. Gemini Pro cache rates follow the same prompt-size tier as regular Gemini Pro input pricing.
- Anthropic Claude uses explicit prompt caching (
cache_control). Cache write tokens are billed at 1.25x input price, cache read tokens at 0.10x input price. Both token types are reported in usage records. - Models without configured cache prices are billed at regular input price even if a request includes cache-like metadata.
Usage records may include cached_tokens and cache_creation_input_tokens when those fields are available from the provider.
Auto-Routing
If you do not specify a model, AIOrouter can select a provider based on availability, model capability, context fit, and current routing policy. You can see which provider handled a request in the X-Provider response header.
Current Exchange Rate
| Date | Source | USD → CAD Rate | 2% Buffer Applied |
|---|---|---|---|
| 2026-06-20 | Estimated (estimated — live Bank of Canada rate unavailable) | 1.38 | Yes |
Token costs in CAD are computed as: USD_price × CAD_USD_rate. The rate above is refreshed daily
at 09:00 EST from the Bank of Canada VALET API with a 2% buffer applied per pricing policy.
Official Pricing Sources
- Anthropic Claude pricing: Anthropic API pricing
- Google Gemini pricing: Google Gemini API pricing documentation
- DeepSeek pricing: DeepSeek API pricing documentation
- Alibaba Qwen and GLM pricing: Alibaba Model Studio Console
- Kimi pricing: Moonshot public API documentation
Data freshness: This BETA catalog was updated on 2026-06-20. Token rates are public provider rates. Claude Fable 5 service suspended (U.S. Commerce Dept export controls effective 2026-06-13 — Anthropic blocked all external Fable 5/Mythos 5 access). Fallback: use Claude Opus 4.8 for equivalent capabilities.