Models and pricing
Review the models that Engy serves, their checkpoints and capabilities, and their per-token prices.
This page lists the models that you can use through the Engy API, along with their capabilities and prices. It is intended for developers who are choosing a model or estimating costs.
Models
Specify the model ID in the model parameter of your request.
| Model ID | Checkpoint | Context window (tokens) | Input | Zero data retention |
|---|---|---|---|---|
deepseek-v4-flash-0731 | deepseek-ai/DeepSeek-V4-Flash-0731 | 1,048,576 | Text | Yes |
deepseek-v4.1-flash | deepseek-ai/DeepSeek-V4.1-Flash | 1,048,576 | Text, image | Yes |
deepseek-v4.1-flash-1017 | deepseek-ai/DeepSeek-V4.1-Flash | 1,048,576 | Text, image | Yes |
glm-5.2 | zai-org/GLM-5.2 | 262,144 | Text | Yes |
glm-5.3 | zai-org/GLM-5.3 | 327,680 | Text | Yes |
glm-5.3-flash | zai-org/GLM-5.3-Flash | 327,680 | Text, image | Yes |
kimi-k3 | moonshotai/Kimi-K3 | 1,113,088 | Text, image | Yes |
qwen3.6-35b-a3b | Qwen/Qwen3.6-35B-A3B-FP8 | 208,192 | Text | No |
qwen3.8-27b | Qwen/Qwen3.8-27B | 1,001,536 | Text, image | Yes |
All models in this table produce text output. deepseek-v4.1-flash-1017 is an alias of deepseek-v4.1-flash. Requests to either ID are served and billed as the same model.
The checkpoint is the Hugging Face repository from which Engy serves the model. Each model is served from exactly one checkpoint. To verify a response, load this checkpoint. For details, see Verify responses.
Models marked Yes in the zero data retention column have ZDR-certified workers. To restrict a request to these workers, set provider.zdr to true. For details, see Data retention.
Pricing
Engy charges per token, with no subscriptions or minimum spend. Prices are in USD per 1 million tokens.
| Model ID | Input | Cached input | Output |
|---|---|---|---|
deepseek-v4-flash-0731 | $0.045 | $0.009 | $0.09 |
deepseek-v4.1-flash | $0.04 | $0.008 | $0.08 |
deepseek-v4.1-flash-1017 | $0.04 | $0.008 | $0.08 |
glm-5.2 | $0.68 | $0.18 | $1.50 |
glm-5.3 | $0.98 | $0.18 | $3.08 |
glm-5.3-flash | $0.135 | $0.027 | $0.45 |
kimi-k3 | $1.95 | $0.195 | $9.75 |
qwen3.6-35b-a3b | $0.045 | $0.015 | $0.30 |
qwen3.8-27b | $0.045 | $0.015 | $0.32 |
Prices can change. For current prices, see the pricing page or query the models endpoint.
Cached input
When a request repeats a prompt prefix that the server has already processed, Engy bills those tokens at the cached input rate. Caching is automatic and requires no configuration.
The number of cached tokens for each request is reported in usage.prompt_tokens_details.cached_tokens.
Video models
| Model ID | Checkpoint | Input | Output |
|---|---|---|---|
MiniMax-H3 | MiniMaxAI/MiniMax-H3 | Text, start and end frames | MP4 video |
Video models are billed per second of video produced. For the current rate, see the pricing page. For usage, see Video generation.
List models with the API
To retrieve the current model list, including capabilities and prices, call the models endpoint:
curl https://api.engy.ai/v1/modelsEach model object contains the following fields:
| Field | Description |
|---|---|
id | The model ID to use in requests. |
input_modalities | The input types that the model accepts, such as text and image. |
output_modalities | The output types that the model produces. |
context_length | The maximum number of tokens in the context window. |
pricing.prompt | The price per input token, in USD. |
pricing.input_cache_read | The price per cached input token, in USD. |
pricing.completion | The price per output token, in USD. |