Merged Qwen3.6 Chat Template Combining Froggeric and Allanchan339 Enhancements
Merge the froggeric and allanchan339 Qwen 3.6 chat templates into a single file, then load it into your llama‑server or Qwen 3.6 35B A3B instance to benefit from strict tool rules, developer role support, JSON‑parsed tool args, and auto‑closing think tags.
Patch your Qwen 3.6 deployment to use the merged template from the provided gist, then run a quick test of a tool‑calling conversation to confirm the new strict rules and JSON parsing work as expected.
Summary
A Reddit user merged two community‑created chat template fixes for Qwen 3.6, one from froggeric and another from allanchan339, to produce a single template that incorporates the best features of both. The merged template adds long strict tool rules with a follow‑up example, accepts a developer role, enables think_on mode, hides historical reasoning by default, parses string tool arguments as JSON, escapes non‑ASCII characters, recognizes </thinking> tags, auto‑closes unclosed tag that froggeric added. The author notes that the merged template is ready for use in local LLaMA deployments and encourages the community to adopt it. The gist link and Reddit discussion provide further details and allow contributors to test or modify the template.
This consolidation simplifies template management for Qwen 3.6 users and removes the need to choose between two competing fixes.
Key changes
- Long strict tool rules with follow‑up example added.
- Developer role accepted.
- think_on mode enabled (replaces think_off).
- Historical reasoning hidden by default.
- String tool arguments parsed as JSON into <parameter> blocks.
- Non‑ASCII characters escaped as uXXXX.
- </thinking> tag recognized.
- Auto‑close unclosed <think> tags before <tool_call>.