Models and pricing
The model served on the AI endpoint, its context window, and the price per million tokens.
Available models
| Model ID | Name | Context window | Input / 1M tokens | Output / 1M tokens | Best for |
|---|---|---|---|---|---|
qwen2.5-7b-instruct | Qwen2.5 7B Instruct | 32k | $0.10 | $0.30 | Arabic-first general assistant, tool calling |
Pass the Model ID as the model field in your requests. Prices are billed
per token from your prepaid balance. See billing and usage.
Capabilities
- Chat completions, streaming and non-streaming — the OpenAI
POST /chat/completionsshape. See the quickstart. - Tool / function calling — pass
toolsand the model returnstool_callsfor your code to execute. Use it non-streaming (set"stream": false); see Call a tool. - Arabic-first — strong in both Arabic and English.
Not yet available: embeddings, vision, and the legacy /completions endpoint
(they return 404).
Query the live list
The live model list is always GET /v1/models on the AI endpoint. Query it
instead of hardcoding assumptions about what is available.
List models from the API
The models endpoint uses the same sk-alwd-… key auth as chat completions:
curl https://ai.alawadi.cloud/v1/models \
-H "Authorization: Bearer $ALAWADI_AI_KEY"{
"object": "list",
"data": [
{
"id": "qwen2.5-7b-instruct",
"object": "model",
"created": 1765432100,
"owned_by": "alawadi"
}
]
}Requesting a model that is not in this list returns 404 with code
model_not_found.
Context window
The model serves a 32k-token context window (prompt plus completion). Keep total tokens under the cap; a request that exceeds it is rejected. Long prompts use more GPU memory and may wait briefly behind other requests on the shared service — see limits and errors for how capacity is shared.