Fast Qwen3.7 vision-language model for text, image, video, tool use, and agentic tasks, with implicit caching and a 1M token context.
Alibaba CloudGeração de Texto1M contextoLançado 15 de jul. de 2026SingaporeEndpoint proprietárioNovoFast Qwen3.7 vision-language model for text, image, video, tool use, and agentic tasks, with implicit caching and a 1M token context.
Supports text, image, and video inputs, thinking mode with enable_thinking and thinking_budget up to 131072 tokens, function calling, structured output, web search, web extractor, code interpreter, image search tools, prefix continuation, and implicit context caching. Web Search and image-search tools are billed per invoked call. Thinking tokens are billed as output tokens.
Também conhecido como Alibaba Cloud Qwen3.7 Flash, Qwen3.7-Flash, qwen3-7-flash
qwen3-7-flash/v1/chat/completionsPOST/v1/responsesPOST/v1/messagesPOST/v1beta/models/qwen3-7-flash:generateContentqwen3.7-flashTarifas pay-as-you-go ao vivo do catálogo EmpirioLabs. Você paga só pelo que usa, sem mínimo mensal.
Qwen3.7 Flash atende a API Chat Completions compatível com OpenAI. Aponte qualquer SDK OpenAI para https://api.empiriolabs.ai/v1 com sua chave de API EmpirioLabs e use o id de modelo qwen3-7-flash. Obtenha uma chave de API no painel EmpirioLabs.
curl https://api.empiriolabs.ai/v1/chat/completions \
-H "Authorization: Bearer $EMPIRIOLABS_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "qwen3-7-flash",
"messages": [
{"role": "user", "content": "Write a haiku about the ocean."}
]
}'from openai import OpenAI
client = OpenAI(
base_url="https://api.empiriolabs.ai/v1",
api_key="YOUR_EMPIRIOLABS_API_KEY",
)
response = client.chat.completions.create(
model="qwen3-7-flash",
messages=[{"role": "user", "content": "Write a haiku about the ocean."}],
)
print(response.choices[0].message.content)Parâmetros de requisição suportados pela API Qwen3.7 Flash na EmpirioLabs. Os padrões valem quando um campo é omitido.
| Parâmetro | Tipo | Padrão | Intervalo / valores | Descrição |
|---|---|---|---|---|
| temperature | number | 0.7 | 0 a 2 | Sampling temperature. 0 is deterministic and 2 is maximum randomness. |
| top_p | number | 0.9 | 0 a 1 | Nucleus sampling probability mass. Lower values make outputs more focused. |
| max_tokens | number | 4096 | 1 a 65536 | Maximum output tokens. |
| stop | string | - | - | Up to 4 strings where the model will stop generating further tokens. |
| enable_thinking | boolean | true | - | Enable reasoning before answering. |
| reasoning_effort | enum | medium | none, low, medium, high, max | Reasoning effort level. none disables thinking. low, medium, high, and max set bounded thinking budgets sized to the selected model. |
| thinking_budget | number | 32768 | 1 a 131072 | Maximum tokens reserved for reasoning when thinking is enabled. |
| vl_high_resolution_images | boolean | true | - | Use higher resolution processing for image inputs. |
| max_pixels | number | 2621440 | 4096 a 16777216 | Maximum pixel count per image when high resolution processing is disabled. |
| video_fps | number | 2 | 0.1 a 10 | Frames per second to sample from video inputs. |
| treat_images_as_video | boolean | false | - | Treat a sequence of images as video frames. |
| tool_web_search | boolean | true | - | Search the web for real-time information. Adds $0.03 to the request cost for each invoked call. |
| tool_web_extractor | boolean | true | - | Extract and read content from URLs. Requires Web Search and Thinking. |
| tool_code_interpreter | boolean | true | - | Run Python code in a sandbox. Requires Thinking. |
Token pricing is tiered by prompt size and steps up above 32K and again above 256K input tokens. Cached prompt tokens are billed at the implicit cache rate for the matching tier. Web Search, Text-to-Image Search, and Image-to-Image Search are billed only when invoked.
Text-to-Image Search and Image-to-Image Search use the Image Search pricing row. Thinking tokens are billed as output tokens.
When this model invokes built-in tools inside a single request, the response carries a normalized usage.tool_usage map alongside the token counts. Tool counts are already factored into cost_usd and are surfaced for transparency.
:variant1
China pricing is discounted versus Singapore. Token pricing is tiered by prompt size and steps up above 32K and again above 256K input tokens. Implicit cache input uses the cached-token row. Web Search, Text-to-Image Search, and Image-to-Image Search are billed only when invoked.
Text-to-Image Search and Image-to-Image Search use the Image Search pricing row. China paid tool calls are $0.01 each. Thinking tokens are billed as output tokens.
Variantes são versões alternativas de Qwen3.7 Flash com id de modelo próprio. Dependendo da variante, podem diferir em região de serviço, preços ou parâmetros suportados; todo o resto funciona igual.
qwen3-7-flash:variant1Na EmpirioLabs, Qwen3.7 Flash é cobrado por uso. A tabela de tarifas ao vivo desta página sempre corresponde ao que a API cobra.
Qwen3.7 Flash suporta uma janela de contexto de 1M tokens com até 65.536 tokens de saída por resposta.
Sim. Qwen3.7 Flash atende a API Chat Completions compatível com OpenAI, então SDKs OpenAI existentes funcionam apontando base_url para https://api.empiriolabs.ai/v1 e definindo o id de modelo qwen3-7-flash.
Qwen3.7 Flash está disponível como 2 ids de modelo: o padrão qwen3-7-flash mais qwen3-7-flash:variant1 (China). Variantes podem diferir em região de serviço, preços ou parâmetros suportados; as tabelas de tarifas de cada uma estão nesta página.
Sim. O playground da EmpirioLabs executa Qwen3.7 Flash no navegador com os mesmos parâmetros que a API expõe, para você testar prompts antes de escrever código.
Crie uma conta EmpirioLabs e gere uma chave em API Keys no painel. A cobrança usa créditos pay-as-you-go, então você paga apenas pelas requisições que faz.
Confira nossos preços ou entre em contato se quiser que seu próprio modelo seja implementado em nossa pilha.
