← Directory

Qwen · CLI

Qwen Code

Open coding agent for terminal workflows.

Official documentation ↗ · Official repository ↗

Identity & surface Documented
CLI Inspect source ↗
Release & license Evidence gap
Latest repository release observed: sdk-typescript-v0.1.18 · published 2026-10-05. Apache-2.0. This may be an SDK, plugin or other surface; it is not a certified CLI version. Inspect source ↗
Install & operating systems Evidence gap
Use the current official installation guide. Commands and OS support are not yet version-certified. Inspect source ↗
Authentication & billing Documented
Native login/API routes require separate eligibility and billing checks. Inspect source ↗
Model compatibility Documented
OpenRouter and OpenAI-compatible endpoints documented. Qwen branding does not restrict this agent to Qwen models; exact model/harness versions still need runtime verification. Inspect source ↗
Reasoning controls Evidence gap
Provider-native model configuration; exact effort mapping not certified. Inspect source ↗
Usage accounting Evidence gap
Token export completeness and provider-reported cost accounting have not yet been certified. Inspect source ↗
Tools & execution Evidence gap
Permissions, sandbox, headless output, MCP/ACP and tool overhead require a version-pinned audit. Inspect source ↗
Plans & allowances Evidence gap
Actual subscription allowance consumption: not quantified in this catalog. No API-price conversion is assumed. Inspect source ↗
Benchmarks & limitations Evidence gap
No original benchmark runs published. AEI, VTE, success and speed are unavailable.

Linked API model examples

Documented route integration permits configuration; these are priced examples, not a tested recommendation for every model, tool, or effort.

Direct API integration source ↗ OpenRouter integration source ↗

What this profile establishes

The overview source was checked on 2026-10-08. This date does not certify every field, model combination or benchmark. Each evidence gap above remains unresolved. A provider integration is not proof that an exact model and effort will work in every agent version or account.

Suggest a correction · How the numbers work