How to read the rate sheetLink to this section
Four rules keep prices and limits unambiguous.
- Prices are per million tokens, in US dollars. Input, cached input, and output are separate rates. A dash means the model does not publish that rate.
- The catalog row is the price. What you see for a model id is what a request to that id is metered at — there is no second, hidden rate for the same model.
- Usage comes from Responses. Token counts are returned in
usageon every response; the public usage series is the traffic the platform actually metered in the last seven days. - Limits are published where they exist. Context window and max output are catalog fields. Anything a model does not publish shows as “not published” rather than a guessed number.
Live pricesLink to this section
Read when this page rendered. Prices are USD per million tokens; context and max output are the catalog’s published limits.
| Model | Input / M | Cached in / M | Output / M | Context | Max output | Data posture |
|---|---|---|---|---|---|---|
| DeepSeek: DeepSeek V4.1 Flashdeepseek/deepseek-v4.1-flash | $0.4725 | $0.0504 | $1.89 | 1.0M | 66K | Training on · No ZDR |
| Meta: Muse Spark 1.3 Contributormeta/muse-spark-1.3-contributor | $0.126 | $0.0126 | $0.252 | 1.05M | — | No training · Zero retention |
| deepseek/deepseek-flashdeepseek/deepseek-flash | $0.18 | $0.018 | $0.72 | 1.0M | — | No training · Zero retention |
| google/gemini-3.8-flashgoogle/gemini-3.8-flash | $0.9 | $0.09 | $4.50 | 1.0M | — | No training · Zero retention |
| x-ai/grok-4.6x-ai/grok-4.6 | $2.00 | $0.2 | $6.00 | 131K | — | No training · Zero retention |
| qwen/qwen3.8-flashqwen/qwen3.8-flash | $0.189 | $0.02016 | $0.5922 | 1.0M | — | Training on · No ZDR |
| Qwen: Qwen3.7 Flashqwen/qwen3.7-flash | $0.0378 | $0.00756 | $0.1638 | 1.0M | 66K | No training · Zero retention |
| z-ai/glm-5.3z-ai/glm-5.3 | $1.40 | $0.14 | $4.40 | 1.0M | — | No training · No ZDR |
| DeepSeek: DeepSeek V4 Flash 0731deepseek/deepseek-v4-flash | $0.5544 | $0.0882 | $1.663 | 1.0M | 66K | Training on · No ZDR |
| z-ai/glm-5.3-flashz-ai/glm-5.3-flash | $0.2268 | $0.063 | $0.756 | 1.0M | — | Training on · No ZDR |
| OpenAI: GPT-5.6 Lunaopenai/gpt-5.6-luna | $0.504 | $0.0504 | $3.024 | 400K | 128K | Training on · No ZDR |
| Meta: Muse Spark 1.3meta/muse-spark-1.3 | $1.25 | $0.125 | $4.25 | 1.05M | — | Training on · No ZDR |
| qwen/qwen3.8-27bqwen/qwen3.8-27b | $0.567 | $0.063 | $4.032 | 1.0M | — | Training on · No ZDR |
| google/gemini-3.7-flashgoogle/gemini-3.7-flash | $0.75 | $0.075 | $3.75 | 1.0M | — | No training · Zero retention |
| meta/muse-glimmer-30bmeta/muse-glimmer-30b | $0.441 | $0.0504 | $1.89 | 128K | — | No training · Zero retention |
| Meta: Muse Spark 1.2meta/muse-spark-1.2 | $1.25 | $0.125 | $4.25 | 1.05M | — | Training on · No ZDR |
| OpenAI: GPT-5.6 Luna Proopenai/gpt-5.6-luna-pro | $0.504 | $0.0504 | $3.024 | 400K | 128K | Training on · No ZDR |
| Qwen: Qwen3.7 Plusqwen/qwen3.7-plus | $0.5 | $0.08064 | $3.00 | 1.0M | — | Training on · No ZDR |
| MiniMax: MiniMax M3minimax/minimax-m3 | $0.378 | $0.0756 | $1.512 | 200K | 512K | Training on · No ZDR |
| DeepSeek: DeepSeek V4 Prodeepseek/deepseek-v4-pro | $1.663 | $0.1663 | $4.99 | 128K | — | Training on · No ZDR |
| Google: Gemma 4 31Bgoogle/gemma-4-31b-it | $0.189 | $0.0756 | $0.504 | 128K | 262K | No training · Zero retention |
| deepseek/deepseek-v4-flash-vision-expdeepseek/deepseek-v4-flash-vision-exp | $0.5544 | $0.05544 | $1.663 | 1.0M | — | Training on · No ZDR |
| OpenAI: GPT-5.6 Solopenai/gpt-5.6-sol | $4.00 | $0.4 | $20.00 | 400K | 128K | No training · Zero retention |
| OpenAI: GPT-5.6 Terraopenai/gpt-5.6-terra | $2.00 | $0.2 | $12.00 | 400K | 128K | No training · Zero retention |
| Qwen: Qwen3.6 35B A3Bqwen/qwen3.6-35b-a3b | $0.248 | $0.063 | $1.485 | 1.0M | — | No training · Zero retention |
| OpenAI: GPT-5.5openai/gpt-5.5 | $5.00 | $0.5 | $30.00 | 400K | 128K | No training · Zero retention |
| Google: Gemma 4 26B A4B google/gemma-4-26b-a4b-it | $0.1764 | $0.063 | $0.504 | 128K | 262K | No training · Zero retention |
Dashes print where a model does not publish a number. Model names link to a full page with that model’s usage series and a ready-to-copy request.
Listing modelsLink to this section
Two catalog doors: one for your key, one public. Both serve the sell-side catalog — prices and limits, never internal cost.
GET https://api.models.sylphx.ai/v1/modelsRequires your key
curl https://api.models.sylphx.ai/v1/models \
-H "Authorization: Bearer $SYLPHX_API_KEY"
# One model document:
curl https://api.models.sylphx.ai/v1/models/openai/gpt-5.5 \
-H "Authorization: Bearer $SYLPHX_API_KEY"
GET https://api.models.sylphx.ai/v1/public/modelsNo key required
curl https://api.models.sylphx.ai/v1/public/models
# One public model, including its 14-day usage series:
curl https://api.models.sylphx.ai/v1/public/models/openai/gpt-5.5
| Field | Type | What it does |
|---|---|---|
id | string | The exact string you send as model. Ids are stable product names. |
display_name | string | Human label for pickers and tables. |
pricing | object | input, output, cache_read, currency, and unit (per_million_tokens). |
limits | object | context_window_tokens and max_output_tokens (null when not published). |
data_policy | object | train and zdr booleans, published per model. |
platform_usage | object | Requests, input/output tokens, and approximate tokens per minute over the last seven days. |
Switching modelsLink to this section
Because model is a field, a switch is a one-line change and a new price. The request and stream shape stay identical.
Only the model id changes
{
"model": "qwen/qwen3.8-flash",
"input": "Summarise this incident report in three bullets."
}
Read the new model’s row first: context window, max output, and data posture can differ. Conversations that used provider-sealed state should stay on the family that produced it — see switching models safely.
Cost mathLink to this section
Cost is the token counts from usage multiplied by the catalog rates. Cached input is metered at the cached rate when the model reports cached tokens.
Per-request cost
cost =
usage.input_tokens / 1_000_000 * price.input
+ usage.cached_tokens / 1_000_000 * price.cache_read # when published
+ usage.output_tokens / 1_000_000 * price.output
NoteWorked example — qwen/qwen3.7-flash
1,000,000 input tokens at $0.0378 plus 250,000 output tokens at $0.1638 = $0.0788 on the current live prices. Recompute with your own mix — the pricing page has an estimator.
Blended averages across models are easy to misread: compare one model at a time, and always state the input/output mix you assumed.
Data postureLink to this section
Each catalog row publishes what the serving provider declares for that model: whether API data may be used for training, and whether zero data retention applies.
- Training on / No training — the provider’s declared handling of API requests for this model.
- Zero retention / No ZDR — whether zero-data-retention applies to requests served by this model.
- The posture is a property of the model row, not of your key, so switching models can change it. Check the row before you move a sensitive workload.
- Platform retention for stored responses and files is separate and described in the privacy policy.