Bug Description

When usingNVIDIA Build provider with model z-ai/glm4.7 , the API returns Thinking Budget Exhausted error, The model ID works on OpenClaw, but Hermes cannot callit successfully.

Steps to Reproduce

1. Run hermes setup and configure custom with NVIDIA Build API key

2. Select model z-ai/glm4.7

3. Try to chat: hernes chat or via gateway

4. Error: ⚠️ Response truncated (finish_reason='length') - model hit max output tokens
   💭 Reasoning exhausted the output token budget — no visible response was produced.
   ─ ⚕ Hermes ─────────────────────────────────────────────────────────────

   ⚠️ Thinking Budget Exhausted

   The model used all its output tokens on reasoning and had none left
   for the actual response.

   To fix this:
   → Lower reasoning effort: /thinkon low or /thinkon minimal
   → Increase the output token limit: set model.max_tokens in
   config.yaml
   5.return to Ollama（cloud） model minimax-m2.7 ，changing nothing，it works

Expected Behavior

⚠️ Response truncated (finish_reason='length') - model hit max output tokens
💭 Reasoning exhausted the output token budget — no visible response was produced.
─ ⚕ Hermes ─────────────────────────────────────────────────────────────

 ⚠️ **Thinking Budget Exhausted**                                         
                                                                          
 The model used all its output tokens on reasoning and had none left      
 for the actual response.                                                 
                                                                          
 To fix this:                                                             
 → Lower reasoning effort: `/thinkon low` or `/thinkon minimal`           
 → Increase the output token limit: set `model.max_tokens` in             
 config.yaml

Actual Behavior

Configuration (configyaml, env, hermes setup)

Affected Component

Setup / Installation

Messaging Platform (if gateway-related)

No response

Operating System

docker in synology（DSM 7.2.2-72806） docker

Python Version

3.13.5

Hermes Version

Hermes Agent v0.8.0 (2026.4.8)

Relevant Logs / Traceback

⚠️  Response truncated (finish_reason='length') - model hit max output tokens
  💭 Reasoning exhausted the output token budget — no visible response was produced.
   ─  ⚕ Hermes  ───────────────────────────────────────────────────────────── 
                                                                              
     ⚠️ **Thinking Budget Exhausted**                                         
                                                                              
     The model used all its output tokens on reasoning and had none left      
     for the actual response.                                                 
                                                                              
     To fix this:                                                             
     → Lower reasoning effort: `/thinkon low` or `/thinkon minimal`           
     → Increase the output token limit: set `model.max_tokens` in             
     config.yaml

Root Cause Analysis (optional)

No response

Proposed Fix (optional)

No response

Are you willing to submit a PR for this?

• I'd like to fix this myself and submit a PR