mirrors/hermes-agent
1
0
Fork 0
mirror of https://github.com/NousResearch/hermes-agent.git synced 2026-06-09 08:21:50 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 10

fix(qwen): correct context lengths for qwen3-coder models and send max_tokens to portal

Browse source 
Based on PR #7285 by @kshitijk4poor.

Two bugs affecting Qwen OAuth users:

1. Wrong context window — qwen3-coder-plus showed 128K instead of 1M.
   Added specific entries before the generic qwen catch-all:
   - qwen3-coder-plus: 1,000,000 (corrected from PR's 1,048,576 per
     official Alibaba Cloud docs and OpenRouter)
   - qwen3-coder: 262,144

2. Random stopping — max_tokens was suppressed for Qwen Portal, so the
   server applied its own low default. Reasoning models exhaust that on
   thinking tokens. Now: honor explicit max_tokens, default to 65536
   when unset.

Co-authored-by: kshitijk4poor <82637225+kshitijk4poor@users.noreply.github.com>
This commit is contained in:
kshitijk4poor 2026-04-11 03:29:09 -07:00 • committed by Teknium
parent f459214010
commit af9caec44f
4 changed files with 45 additions and 6 deletions
Download patch file Download diff file Expand all files Collapse all files

12
run_agent.py
View file

@ -5888,8 +5888,16 @@ class AIAgent:
api_kwargs["tools"] = self.tools
if self.max_tokens is not None:
if not self._is_qwen_portal():
api_kwargs.update(self._max_tokens_param(self.max_tokens))
api_kwargs.update(self._max_tokens_param(self.max_tokens))
elif self._is_qwen_portal():
# Qwen Portal defaults to a very low max_tokens when omitted.
# Reasoning models (qwen3-coder-plus) exhaust that budget on
# thinking tokens alone, causing the portal to return
# finish_reason="stop" with truncated output — the agent sees
# this as an intentional stop and exits the loop. Send 65536
# (the documented max output for qwen3-coder models) so the
# model has adequate output budget for tool calls.
api_kwargs.update(self._max_tokens_param(65536))
elif (self._is_openrouter_url() or "nousresearch" in self._base_url_lower) and "claude" in (self.model or "").lower():
# OpenRouter and Nous Portal translate requests to Anthropic's
# Messages API, which requires max_tokens as a mandatory field.