Phala API pricing
Live prices for every model Phala serves: its own per-1M-token prices, or per image, second or character, next to the cheapest price elsewhere.
Phala serves 18 models with live API prices on SF·SIX, 18 models priced per 1M tokens. The cheapest is Qwen2.5 7B Instruct at $0.10 per 1M input and $0.20 per 1M output tokens. The most used on OpenRouter is DeepSeek V4.1 Flash (5.41T tokens on 2026-10-16).
- Models with live prices
- 18
- Cheapest language model
- Qwen2.5 7B Instruct$0.10 / $0.20 per 1M tokens
- Most used on OpenRouter (2026-10-16)
- DeepSeek V4.1 Flash5.41T tokens
- Newest model
- DeepSeek V4.1 Flash2026-09-10
- Cheapest standard host for
- 1 model of 18
Phala token prices (USD per 1M tokens)
| Model | Input / 1M | Output / 1M | Terms | Cheapest standard price anywhere | Context (tokens) | OpenRouter tokens / day |
|---|---|---|---|---|---|---|
| GLM 5.3 | $0.84 | $2.64 | Standard | $0.50 / $1.51Sail Research (fp8) | 1,048,576 | 942.62B |
| DeepSeek V4.1 Flash | $0.21 | $0.84 | Standard | $0.09 / $0.18Decart (fp4) | 1,048,576 | 5.41T |
| GLM 5.3 Flash | $0.1275 | $0.425 | Standard · nvfp4 | $0.075 / $0.25DeepInfra (fp4) | 1,048,576 | 1.73T |
| GLM 5.2 | $1.26 | $3.00 | Standard · fp8 | $0.189 / $0.594Baidu (fp8) | 1,048,576 | 140.87B |
| DeepSeek V4 Flash 0731 | $0.308 | $0.924 | Standard | $0.0598 / $0.1795Baidu (fp8) | 1,048,576 | 620.6B |
| DeepSeek V4 Pro 0813 | $0.957 | $2.88 | Standard | $0.1333 / $0.40Baidu (fp8) | 1,048,576 | 161.08B |
| gpt-oss-120b | $0.15 | $0.60 | Standard | $0.03 / $0.15Near AI (fp4) | 131,072 | 86.23B |
| Qwen3.8 27B | $0.20 | $2.50 | Standard | $0.02 / $0.50Tenstorrent (fp4) | 1,000,000 | 62.19B |
| Kimi K3 | $2.55 | $12.75 | Standard | $0.504 / $8.10Sail Research (fp4) | 1,048,576 | 203.41B |
| Kimi K2.6 | $1.09 | $4.60 | Standard | $0.4323 / $1.82Baidu (fp4) | 262,144 | Not in named rows |
| GLM 5.1 | $1.21 | $4.20 | Standard | $0.9646 / $3.03Baidu (fp8) | 204,800 | Not in named rows |
| Kimi K2.7 Code | $0.95 | $4.00 | Standard | $0.7125 / $3.00StreamLake | 262,144 | Not in named rows |
| DeepSeek V3.2 | $1.00 | $1.00 | Standard | $0.2088 / $0.3096GMI Cloud (fp8) | 163,840 | Not in named rows |
| Qwen3.6 35B A3B | $0.20 | $1.27 | Standard | $0.05 / $0.70Darkbloom (fp4) | 262,144 | 386.14B |
| Qwen3.6 27B | $0.32 | $3.25 | Standard | $0.25 / $2.00Chutes (fp8) | 262,144 | Not in named rows |
| Qwen3.5-27B | $0.30 | $2.40 | Standard | $0.195 / $1.56Alibaba | 262,144 | Not in named rows |
| Hy3 | $0.15 | $0.64 | Standard | $0.132 / $0.528Tencent (fp8) | 262,144 | 477.68B |
| Qwen2.5 7B Instruct | $0.10 | $0.20 | Standard | Phalacheapest standard host | 32,768 | Not in named rows |
Phala’s own endpoint for each model (standard tier and default region first), next to the cheapest standard price from any provider. Most widely served first.
Method
Prices are USD per 1M tokens at the standard service tier in the default region: the cheapest qualifying provider per model, with flex, batch, priority and fast endpoints and unusually low outliers never used as the headline. “Blended” weights input and output 3:1. Missing prices are unknown, not free. OpenRouter tokens per day are the latest full UTC day of OpenRouter’s public rankings (named rows only). Prices move: see AI API price changes for recent cuts and increases.