Hermes Agent works with 200+ models, so the best choice balances agentic ability, long-running reliability, and cost — because its autonomous loops make a lot of calls. Claude Sonnet 4.6 is the strongest all-round driver; for value, open-weight models like DeepSeek V4 Pro, MiniMax M3, and GLM 5.2 deliver near-frontier results at a fraction of the price, which matters for an agent that runs unattended.
The best all-round driver — strong tool use and reliability for long autonomous runs, without the flagship price.
The value champion: near-frontier quality at roughly a tenth of the cost, and open weights you can self-host — ideal for an agent that runs constantly.
Built for agents — dependable tool use and a 1M context at low cost, well suited to Hermes's long-horizon tasks.
Strong agentic coding at open-weight prices when Hermes is doing development work.
Fast with a 1M context and a great price — keeps long, tool-heavy runs cheap and snappy.
A capable open-weight flagship with a 1M context for the harder autonomous tasks.
The model picks the moves; the agent runs the loop, the tools, and the guardrails. Once you've chosen a model, see which agent gets the most out of it.