Qwen 3.6 Plus Preview — Public model availability¶

Development line: project:qwen-3-6-plus-preview · thread public-model-availability
Last event: 2026-03-31 · 1 dated since 2026-03-31 · Researched: 2026-09-05 · confidence: medium

What it is¶

Qwen 3.6 Plus Preview is a Qwen-hosted preview model available as qwen/qwen3.6-plus-preview:free on OpenRouter.

• Agentic coding, front-end work, and general reasoning.
• 1,000,000-token context window.
• OpenRouter lists it as free, but prompts and completions may be used to improve the model.

Development line¶

• 2026-03-31 — OpenRouter availability signal for Qwen 3.6 Plus Preview. On 2026-03-31, the record linked Qwen 3.6 Plus Preview to an OpenRouter URL marked as a free endpoint. This proposes recording that dated public-availability signal, rather than a verified model launch: the supplied evidence does not establish the operator, capabilities, terms, or whether the endpoint was live on that date.

What changed¶

2026-03-31 — Qwen 3.6 Plus Preview became available through OpenRouter as a free preview endpoint; OpenRouter identifies the release date as 2026-03-30. 2026-04-03 — an unresolved community report said the model was unavailable through the international DashScope API, despite availability on Qwen Chat and the free Qwen Code CLI.

How to use this¶

From 2026-03-31, practitioners should treat Qwen 3.6 Plus Preview as a possible OpenRouter free-endpoint option only after checking current availability and terms; this record establishes no capability or provenance claims.

1. Create an OpenRouter API key, then send a standard OpenAI-compatible request to POST /api/v1/chat/completions with model qwen/qwen3.6-plus-preview:free. — https://openrouter.ai/docs/api_reference/overview
2. Start with a bounded evaluation task such as repository review or long-document analysis; do not submit confidential material because the model page says prompts and completions may be collected for improvement. — https://openrouter.ai/qwen/qwen3.6-plus-preview:free
3. For quota failures, inspect the key endpoint and honor Retry-After; use exponential backoff for 429 responses. — https://openrouter.ai/docs/api_reference/limits

Best practices¶

• Treat this as a preview evaluation path: pin the exact model ID and retain outputs from representative tests before making it a dependency. — https://openrouter.ai/qwen/qwen3.6-plus-preview:free
• Do not send sensitive prompts or completions through this endpoint because OpenRouter states that they may be collected for model improvement. — https://openrouter.ai/qwen/qwen3.6-plus-preview:free
• Implement backoff and quota monitoring for free-model requests; provider capacity and platform limits can return 429 errors. — https://openrouter.ai/docs/api_reference/limits

Superseded by this¶

• 2026-03-31: the assumption that the release date was definitively March 31 is corrected by OpenRouter’s current model page, which lists March 30, 2026.
• 2026-03-31: treating Qwen 3.6 Plus Preview as a local or open-weight Qwen release is unsupported; the verified access path is the hosted OpenRouter listing.

Still unknown¶

• No first-party Qwen model card, benchmark report, parameter count, architecture specification, post-preview pricing, or formal availability statement was found for this exact model.
• The April 3 DashScope availability claim is a user report in a Qwen GitHub issue, not a maintainer confirmation.
• The current OpenRouter page remains accessible and labels the model free, but a public page alone does not prove capacity, provider routing, or suitability for production traffic.

Sources¶

Agent brief¶

• Subject: project:qwen-3-6-plus-preview, thread public-model-availability, 1 dated events 2026-03-31 → 2026-03-31.
• Practical note: From 2026-03-31, practitioners should treat Qwen 3.6 Plus Preview as a possible OpenRouter free-endpoint option only after checking current availability and terms; this record establishes no capability or provenance claims.
• Confidence: medium. Dated supersedes above are the authority for what is obsolete.