Current as of April 2026. GPT-4 is a legacy powerhouse for Hermes Agent that offers high reliability for tool-calling but carries a premium price and a restrictive context window. It serves as a stable choice for complex multi-platform automation where execution precision is more critical than speed or cost.
Specs
| Provider | OpenAI |
| Input cost | $30 / M tokens |
| Output cost | $60 / M tokens |
| Context window | 8K tokens |
| Max output | 4K tokens |
| Parameters | N/A |
| Features | function_calling |
What it’s good at
Precise Tool Execution
This model handles the 47 built-in Hermes tools with high accuracy, rarely failing to format parameters for shell commands or messaging platform APIs.
Identity Stability
It maintains a consistent persona across different platforms like Telegram and Discord, effectively utilizing the closed learning loop to sustain its identity.
Where it falls short
Prohibitive Pricing
At $30 per million input and $60 per million output tokens, it is significantly more expensive than modern frontier models.
Restrictive Context Window
The 8K token limit is a severe bottleneck for Hermes’ persistent memory, leading to frequent truncation during long autonomous runs.
Best use cases with Hermes Agent
- Critical Shell Automation — The high reliability in tool-calling makes it safer for running sensitive terminal commands on local or SSH environments where errors are costly.
- Multi-Platform Logical Bridging — It excels at reasoning through complex instructions that require moving data between different protocols like Slack, WhatsApp, and Discord.
Not ideal for
- Continuous Channel Monitoring — The $30/1M input cost makes it unaffordable to leave Hermes idling while watching high-volume chat channels for triggers.
- Long-Session Memory Retention — The 8K context window fills up quickly when tracking multiple cross-session interactions or extensive tool logs.
Hermes Agent setup
Set the provider to OpenAI and the model ID to openai/gpt-4; ensure your account has sufficient credits to cover the high per-token costs.
Hermes makes custom endpoints easy. Run:
hermes model
Choose Custom endpoint from the menu. Enter the base URL and model identifier when prompted:
- Base URL:
https://api.haimaker.ai/v1 - Model:
openai/gpt-4
Hermes stores the selection and uses it for all subsequent agent runs across whatever platforms you have wired up (Telegram, Discord, Slack, etc.). Tune HERMES_STREAM_READ_TIMEOUT and related env vars if you’re hitting slow providers.
How it compares
- vs GPT-4o — GPT-4o is faster, cheaper at $5/$15 per 1M tokens, and offers a 128K context window compared to GPT-4’s 8K limit.
- vs Claude 3.5 Sonnet — Sonnet provides superior tool-use reasoning and a 200K context window for a fraction of the cost at $3/$15 per 1M tokens.
Bottom line
GPT-4 remains a reliable workhorse for tool-heavy automation, but its 8K context and high pricing make it difficult to justify over modern models like GPT-4o.
For more, see our Hermes local-LLM setup guide.