Qwen 3.7 Flash - Kyma API

Overview

This is a vision-language reasoning model from Alibaba that accepts text, image, and video inputs and returns text. It supports a 1M token context window and outputs up to 8,192 tokens. The model operates in the cheap cost tier and is optimized for high-throughput workloads. On Kyma, it runs through an OpenAI-compatible endpoint with automatic request failover if a serving path degrades. Prompt caching is enabled, billing repeated prefixes at this model’s cached input rate. Responses include the exact request cost in usage.cost and the executed model ID in the X-Kyma-Model header. It does not support structured outputs, so you will need to parse JSON or enforce formatting manually. The 1M context window is large, but the 8,192 token output limit restricts long-form generation.

Specs

Field	Value
Model ID	`qwen3.7-flash`
Best for	Multimodal agents, visual coding and computer use at high volume
Context window	1M
Max output tokens	8K
Input modalities	Text, Image, Video
Output modalities	Text
Tool calling	Yes
Structured outputs	Yes
Prompt caching	Yes
Speed	Fast
Cost band	Cheap
Release stage	Stable

Pricing

	Per 1M tokens
Input	$0.0498
Output	$0.2164

Above a long prompt this model costs more per token, because the infrastructure serving it charges more there. The threshold is on the prompt, so you know which rate applies before you send.

Prompt length	Input	Output
Standard	$0.0498	$0.2164
Above 32,000 tokens	$0.135	$0.54
Above 256,000 tokens	$0.27	$1.08

Use this when

Multimodal agent orchestration — Route text, image, and video inputs to tool-calling agents that run at high volume.
Visual code review — Analyze screenshots, UI mockups, and video walkthroughs to generate code suggestions.
Document analysis at scale — Process large batches of mixed-media files within a 1M token context window.
High-throughput chat routing — Handle conversational workloads with fast response times and low compute overhead.

Not ideal for

Avoid this model when you require guaranteed structured JSON outputs or need to generate responses longer than 8,192 tokens.

Pick something else when

You need guaranteed structured JSON outputs: use gpt-5.6-terra.
You require longer output sequences: use qwen-3.7-max.

Example

from openai import OpenAI

client = OpenAI(base_url="https://kymaapi.com/v1", api_key="kyma-...")

response = client.chat.completions.create(
    model="qwen3.7-flash",
    messages=[{"role": "user", "content": "..."}],
    tools=[...],  # function calling supported
)

Agent query example

Ask the API which models fit, instead of hardcoding an id:

curl "https://kymaapi.com/v1/models?recommended_for=chat&tools=true&quality_tier=efficient"

FAQ

Does this model support function calling? Yes, it supports tool use and reasoning, allowing you to pass tool definitions and parse the model’s calls. How does prompt caching affect billing? Kyma bills cached prompt prefixes at this model’s cached input rate, reducing costs for repeated system prompts or conversation history. What happens if the serving path fails? Kyma automatically reroutes your request to an alternate serving path, ensuring high availability without manual retries.

​Overview

​Specs

​Pricing

​Use this when

​Not ideal for

​Pick something else when

​Example

​Agent query example

​FAQ