agentos.guide › blog

Hermes Agent Bedrock: The Quick-Start, The Config Keys And Where AWS Fits In An Agent OS

By Julian Goldie · 7 October 2026 · agentos.guide

Hermes Agent Bedrock support is native, so you choose "AWS Bedrock" in hermes model and Hermes calls Amazon Bedrock directly with your normal AWS credentials.

You don't need LiteLLM, a proxy or OpenRouter in the middle, because the official Hermes docs list bedrock as a built-in provider.

Below you'll get the quick-start, the real config keys, the models that make sense, the costs I could verify and how Bedrock fits an Agent OS workflow.

What Hermes Agent Bedrock support actually is

Hermes Agent is the open-source AI agent from Nous Research, and it is the harness that runs tools, memory and messaging.

Amazon Bedrock is the AWS service that serves models from many makers through one AWS account.

The official docs say Hermes supports Bedrock as a native provider, with IAM authentication, Guardrails, cross-region inference profiles and Bedrock's foundation models.

Hermes routes each model family through the API that suits it best.

All three routes share the same AWS credentials and the same region setting.

You set it up once and then switch between Claude, Nova and Llama with the /model command.

An honest note before the quick-start

I run Hermes every day inside my Agent OS, but I haven't published a Bedrock test of my own.

Everything here comes from the official Hermes docs, the Hermes source code in my install (v0.21.4) and the AWS docs, all read on 12 October 2026.

I'll tell you where I couldn't verify something.

Hermes Agent Bedrock quick-start

You need four things first.

• You need an AWS account with a valid payment method.
• You need an AWS identity that is allowed to call Bedrock.
• You need a region where Bedrock offers the model you want.
• You need Hermes Agent installed already.

1. Give your AWS identity the right permissions

The Hermes docs list four permissions as the minimum.

• bedrock:InvokeModel and bedrock:InvokeModelWithResponseStream are for inference.
• bedrock:ListFoundationModels and bedrock:ListInferenceProfiles are for model discovery.

You only add bedrock:GetInferenceProfile if your default model is an application inference profile ARN.

2. Check model access

AWS changed this, so older tutorials are out of date.

The AWS model access docs say access to many Bedrock models is enabled by default when you have the right AWS Marketplace permissions.

Anthropic models are the exception, because AWS says first-time customers must submit use case details once per account or organisation.

You submit that form by selecting an Anthropic model in the Bedrock console's model catalogue.

3. Sign the machine in to AWS

Hermes uses the standard boto3 credential chain, so there is no Hermes-specific key.

On a laptop, a named profile is the simplest route.

aws configure --profile hermes

aws sts get-caller-identity --profile hermes

On AWS compute, you attach an IAM role and the docs say Hermes detects it automatically.

You should never paste an AWS secret into a chat with the agent.

4. Install the Bedrock extra

Bedrock needs the AWS library boto3, which Hermes ships as an optional extra.

The docs give this exact command.

cd ~/.hermes/hermes-agent && python -c "import pm; pm.sync_venv(['bedrock'], explicit=True)"

Restart Hermes after it finishes.

5. Choose AWS Bedrock in the model picker

hermes model

Choose "More providers...", then "AWS Bedrock".

The wizard asks for your AWS region and then asks you to choose an authentication method.

The IAM credential chain is option 1 and is marked as recommended.

A Bedrock API Key is option 2, and the wizard describes it as useful when an admin hands keys out to a team.

Hermes then discovers the text models in your region and shows the recommended ones first.

6. Check it and chat

hermes doctor

hermes chat

The docs say hermes doctor checks your AWS credentials, the boto3 install, whether the Bedrock API is reachable and how many models your region has.

You can also try one model for a single session without changing your saved config.

AWS_PROFILE=hermes AWS_REGION=us-east-1 hermes chat --provider bedrock --model us.anthropic.claude-sonnet-4-6

🔥 Want the exact Hermes setup I run inside my Agent OS?

Inside the AI Profit Boardroom you get the installable Agent OS, step-by-step video tutorials, four coaching calls a week and 3,400+ members building real automations.

→ Get access here

The Hermes Agent Bedrock config keys

After the wizard runs, the docs say your ~/.hermes/config.yaml holds three lines that matter.

model.provider: bedrock

model.default: us.anthropic.claude-sonnet-4-6

bedrock.region: us-east-2

The wizard also writes a model.base_url line that points at the Bedrock runtime endpoint for your region.

Hermes resolves the region in a fixed order.

It reads bedrock.region first, then AWS_REGION, then AWS_DEFAULT_REGION, and it falls back to us-east-1.

Guardrails have one catch that the docs spell out.

AWS does not apply Guardrails to the Mantle endpoint that serves the OpenAI GPT-5.x models, so you use a Converse-served model when a guardrail is required.

Which Bedrock models make sense for an agent

The Hermes docs list these IDs, with the recommended ones at the top of the picker.

Bedrock's own catalogue is bigger than that shortlist.

The AWS models at a glance page listed newer Claude models, Mistral Large 3, Qwen3 models, Kimi K3, Grok 4.7 and GLM 5.3 when I read it on 12 October 2026.

Hermes discovers what your own region offers, so trust the picker over any table in a blog post.

I'd start with a Claude model for agent work, because tool calling is the whole job and the Claude route keeps prompt caching.

I'd use a small model like Nova Micro for simple, high-volume jobs where speed and price matter more.

Tool calling and caching caveats

An agent that can't call tools is just a chatbot, so check this before you commit to a model.

• Hermes converts its tool definitions into the Bedrock Converse format for you.
• The Hermes source keeps a short list of models that reject tools, which covers DeepSeek R1, Stability image models and embedding models.
• The source treats every other model as tool-capable, so a small model may accept tools and still use them badly.
• The docs say prompt caching is applied automatically, but only for Claude and Amazon Nova model IDs.
• The docs say models with a us. prefix use cross-region inference, and models with a global. prefix route across all available regions.

That last point matters if you chose Bedrock for data residency, so ask your compliance person to confirm the profile ID you use.

How Hermes Agent Bedrock fits an Agent OS workflow

My Agent OS is a dashboard that runs a crew of agents under one shared memory, and Hermes is one of the workers.

I keep one Hermes profile per model, so each brain has its own clean config.

Bedrock would slot in the same way, as one more profile with bedrock as its provider.

You create a profile with hermes profile create and run any command against it with hermes -p and the profile name.

The fallback chain is the part I'd set up on day one.

The docs list bedrock as a supported fallback provider, and hermes fallback add uses the same picker as hermes model.

That means a rate limit on one provider doesn't stop a scheduled job.

The docs also say Bedrock works with every Hermes gateway platform, so a Bedrock-backed agent can still answer you in Telegram.

If you want the bigger picture, my 3-layer Hermes OS guide shows how the brain, the cockpit and the memory fit together.

🔥 Want the exact Hermes setup I run inside my Agent OS?

Inside the AI Profit Boardroom you get the installable Agent OS, step-by-step video tutorials, four coaching calls a week and 3,400+ members building real automations.

→ Get access here

What Hermes on Bedrock costs

Hermes Agent is free and open source, so the bill is the AWS bill.

Bedrock charges per token on demand, and the price depends on the model, the region and the service tier.

I read the Amazon Bedrock pricing page on 12 October 2026.

The lowest regional prices I could read were $0.50 per million input tokens and $1.50 per million output tokens for Mistral Large 3.

DeepSeek v3.2 showed $0.62 and $1.85 at its lowest, and other regions on the page were higher.

The Claude and Amazon Nova prices load dynamically on that page and I couldn't read them reliably, so I won't quote them.

The page also describes batch inference and service tiers called Standard, Priority, Flex and Reserved.

An agent re-sends its instructions and tool list on every turn, so caching support matters more than the headline price.

Why a business would choose Bedrock

• The model spend lands on the AWS bill the company already manages.
• Access is controlled with IAM roles and policies instead of a loose API key.
• You choose the AWS region the agent calls.
• You can attach Amazon Bedrock Guardrails to model calls.
• AWS's data protection page says model providers don't have access to Bedrock logs or to customer prompts and completions.

I'm not going to tell you Bedrock makes you compliant with anything, because that is your auditor's call.

Troubleshooting Hermes Agent Bedrock errors

The region is behind more problems than people expect.

A model can exist on Bedrock and still not be offered in your region.

The Hermes setup code only shows models your region can route to, so a model missing from the picker usually points at the region.

Run hermes doctor first, because it tells you which layer is broken.

Alternatives if Bedrock is more than you need

Bedrock is the right call for AWS-first teams, and it is more setup than most solo operators need.

The Hermes docs call Nous Portal the recommended way to run Hermes Agent, and hermes setup --portal sets it up in one command.

If you already pay for Claude, my Hermes on a Claude subscription guide shows another route.

If you want free brains to test with, my run Hermes free forever guide covers the ones I've used.

My ranked list of brains is in the best Hermes Agent model guide.

Two sister guides from this batch go deeper on the surrounding pieces: the Hermes Agent backend guide and the explainer on Hermes Agent by Nous Research.

FAQ

Does Hermes Agent support Amazon Bedrock?

Yes. The official Hermes docs say Hermes Agent supports Amazon Bedrock as a native provider. You choose AWS Bedrock in hermes model, and the provider name in config.yaml is bedrock.

Do I need LiteLLM or OpenRouter to use Bedrock with Hermes?

No. Hermes calls Bedrock directly through three routes: the Anthropic SDK's Bedrock client for Claude, Bedrock Mantle for OpenAI GPT-5.x models and the native Converse API for everything else.

Which credentials does Hermes use for Bedrock?

It uses the standard boto3 credential chain, so an IAM role, a named AWS profile, SSO or the usual AWS access key environment variables all work. The wizard also offers a Bedrock API Key option.

Which Bedrock model should I use with Hermes Agent?

The Hermes docs recommend Claude Sonnet 4.6 as the best balance of speed and capability in their list. I'd start with a Claude model for agent work because tool calling matters most and the Claude route keeps prompt caching.

Why do I get an on-demand throughput error?

You used a bare foundation model ID. The docs say to use an inference profile ID with a us. or global. prefix, such as us.anthropic.claude-sonnet-4-6.

Has Julian tested Hermes on Bedrock?

Not in a published test. This guide is based on the official Hermes docs, the Hermes source code and the AWS docs, read on 12 October 2026.

Final word on Hermes Agent Bedrock

Bedrock is a first-class route for Hermes, and the hard part is the AWS setup and not the agent.

Run the six quick-start steps, check it with hermes doctor, and you'll have a working Hermes Agent Bedrock setup.

Related guides on agentos.guide

→ Best Hermes Agent Model

→ Run Hermes On Your Claude Subscription

→ The 3-Layer Hermes OS

→ Run Hermes Free Forever

📺 Video notes + links to the tools 👉

🎥 Learn how I make these videos 👉

🆓 Get a FREE AI Course + Community + 1,000 AI Agents 👉

← all guides on agentos.guide