Amazon Bedrock Mantle Provider for OpenClaw

OpenClaw ships with a built-in Amazon Bedrock Mantle provider that targets the Mantle OpenAI-compatible API. Mantle serves open-source and third-party models (GPT-OSS, Qwen, Kimi, GLM, and others) via a standard /v1/chat/completions interface running on Bedrock infrastructure. Additionally, Mantle offers Anthropic Claude models through an Anthropic Messages endpoint.

Getting started

Pick the authentication approach that fits your setup and complete the configuration steps.

Explicit bearer token

Best for: environments where you already have a Mantle bearer token.

Set the bearer token on the gateway host

export AWS_BEARER_TOKEN_BEDROCK="..."

You can optionally specify a region, which defaults to us-east-1:

export AWS_REGION="us-west-2"

Verify models are discovered

openclaw models list

Models that are discovered show up under the amazon-bedrock-mantle provider. Unless you need to change defaults, no further configuration is necessary.

IAM credentials

Best for: using AWS SDK-compatible credentials (shared config, SSO, web identity, instance or task roles).

Configure AWS credentials on the gateway host

Any AWS SDK-compatible authentication source is acceptable:

export AWS_PROFILE="default"
export AWS_REGION="us-west-2"

Verify models are discovered

openclaw models list

From the credential chain, OpenClaw automatically creates a Mantle bearer token.

Tip

If AWS_BEARER_TOKEN_BEDROCK remains unset, OpenClaw generates the bearer token on your behalf from the AWS default credential chain, which covers shared credentials/config profiles, SSO, web identity, and instance or task roles.

Automatic model discovery

When AWS_BEARER_TOKEN_BEDROCK is provided, OpenClaw uses it as-is. If not, OpenClaw tries to produce a Mantle bearer token from the AWS default credential chain. After that, it queries the region's /v1/models endpoint to find available Mantle models.

To keep the Mantle plugin active while turning off automatic discovery and IAM bearer-token generation, disable the plugin-owned discovery toggle:

openclaw config set plugins.entries.amazon-bedrock-mantle.config.discovery.enabled false

Note

The bearer token is the same AWS_BEARER_TOKEN_BEDROCK used by the standard Amazon Bedrock provider.

Supported regions

us-east-1, us-east-2, us-west-2, ap-northeast-1, ap-south-1, ap-southeast-3, eu-central-1, eu-west-1, eu-west-2, eu-south-1, eu-north-1, sa-east-1.

Manual configuration

If you would rather use explicit configuration than auto-discovery:

{
  models: {
    providers: {
      "amazon-bedrock-mantle": {
        baseUrl: "https://bedrock-mantle.us-east-1.api.aws/v1",
        api: "openai-completions",
        auth: "api-key",
        apiKey: "env:AWS_BEARER_TOKEN_BEDROCK",
        models: [
          {
            id: "gpt-oss-120b",
            name: "GPT-OSS 120B",
            reasoning: true,
            input: ["text"],
            cost: { input: 0, output: 0, cacheRead: 0, cacheWrite: 0 },
            contextWindow: 32000,
            maxTokens: 4096,
          },
        ],
      },
    },
  },
}

An explicit non-empty models list determines membership and overrides discovered rows, including the Claude rows listed below. For rows that match, an explicit input takes precedence; when the source row lacks input, discovery can supply that capability metadata. Leave models out to keep the automatic Mantle catalog, or include the full set of Claude model entries you intend to use.

Advanced configuration

Reasoning support

Reasoning support gets inferred from model IDs that contain patterns such as thinking, reasoner, reasoning, deepseek.r, gpt-oss-120b, or gpt-oss-safeguard-120b. During discovery, OpenClaw sets reasoning: true automatically for models that match.

Endpoint unavailability

Should the Mantle endpoint be unreachable, return no models, or bearer-token resolution fail, discovery yields an empty result and the implicit provider gets skipped. OpenClaw does not raise an error; any other configured providers keep functioning normally.

Claude via the Anthropic Messages route

When automatic discovery manages the model list, OpenClaw adds five Claude models after a successful lookup, no matter what /v1/models returns: amazon-bedrock-mantle/anthropic.claude-opus-5 (Claude Opus 5), amazon-bedrock-mantle/anthropic.claude-sonnet-5 (Claude Sonnet 5), amazon-bedrock-mantle/anthropic.claude-opus-4-7 (Claude Opus 4.7), amazon-bedrock-mantle/anthropic.claude-mythos-5 (Claude Mythos 5), and amazon-bedrock-mantle/anthropic.claude-mythos-preview (Claude Mythos Preview). These rely on the anthropic-messages API surface and transmit through the same bearer-authenticated Anthropic-compatible endpoint (<mantle-base>/anthropic), so the AWS bearer token is not handled like an Anthropic API key.

Claude Opus 5 offers a 1,000,000-token context window, a 128,000-token output cap, image input, and $5/$25 input/output pricing. Adaptive thinking starts at high; /think off turns thinking off, and /think xhigh|max applies the model's native effort levels. Caller-selected sampling parameters are dropped by OpenClaw.

Claude Sonnet 5 always engages adaptive thinking and starts at high effort. /think off and /think minimal are translated to low because the Mantle route cannot turn off thinking. Custom temperature is also excluded from Sonnet 5 requests by OpenClaw.

Claude Mythos 5 is restricted access. It provides a 1,000,000-token context window and a 128,000-token output cap, always uses adaptive thinking, translates /think off and /think minimal into low, and drops caller-selected sampling parameters.

Claude Mythos Preview always asks for reasoning, defaulting to high effort when no /think level is set (mapped from xhigh/max down to high, and minimal up to low). Opus 4.7 on Mantle streams without model-provided reasoning, and OpenClaw drops its temperature parameter since Opus 4.7 does not accept sampling overrides on this route; Mythos Preview accepts a temperature override normally.

An explicit non-empty models.providers["amazon-bedrock-mantle"].models list determines membership and replaces the entire discovered catalog. Matching rows can inherit the discovered input capability only when the source row left it out. Leave the list out when you want these built-in Claude rows.

Relationship to Amazon Bedrock provider

Bedrock Mantle is a distinct provider from the standard Amazon Bedrock provider. Mantle uses an OpenAI-compatible /v1 surface for its OSS catalog, while the standard Bedrock provider relies on the native Bedrock Converse API.

When present, both providers use the same AWS_BEARER_TOKEN_BEDROCK credential.

Related

• Amazon Bedrock, Native Bedrock provider for Anthropic Claude, Titan, and other models.

• Model selection, Choosing providers, model refs, and failover behavior.

• OAuth and auth, Auth details and credential reuse rules.

• Troubleshooting, Common issues and how to resolve them.