~/.openclaw/openclaw.json for Aamzon Bedrock API KEY

openclaw.json.md

cat ~/.openclaw/openclaw.json | jq -r .models

{
  "mode": "merge",
  "providers": {
    "bedrock": {
      "baseUrl": "https://bedrock-runtime.us-east-1.amazonaws.com",
      "apiKey": "${AWS_BEARER_TOKEN_BEDROCK}",
      "api": "bedrock-converse-stream",
      "auth": "api-key",
      "authHeader": true,
      "models": [
        {
          "id": "openai.gpt-oss-120b-1:0",
          "name": "GPT OSS",
          "reasoning": true,
          "input": [
            "text"
          ],
          "cost": {
            "input": 0,
            "output": 0,
            "cacheRead": 0,
            "cacheWrite": 0
          },
          "contextWindow": 262144,
          "maxTokens": 8192
        }
      ]
    }
  }
}

% cat ~/.openclaw/openclaw.json | jq -r .agents.defaults

{
  "model": {
    "primary": "bedrock/openai.gpt-oss-120b-1:0"
  },
  "models": {
    "bedrock/openai.gpt-oss-120b-1:0": {
      "alias": "gpt-oss-120b"
    }
  },
  ...
}

# generate Bedrock API KEY at https://console.aws.amazon.com/bedrock/home#/api-keys
export AWS_BEARER_TOKEN_BEDROCK=xxxxxxxxx
openclaw gateway run --verbose // make sure no errors

Now when you type /new in Telegram, you should see

✅ New session started · model: bedrock/openai.gpt-oss-120b-1:0

Then it works!

Model IDs:

Qwen3 Coder - qwen.qwen3-coder-30b-a3b-v1:0
Clause Haiku 4.5 - us.anthropic.claude-haiku-4-5-20251001-v1:0
GPT OSS 120b - openai.gpt-oss-120b-1:0

Thanks for sharing this, super helpful!

For those outside the US, I had to use: "global.anthropic.claude-haiku-4-5-20251001-v1:0" in order to get it to work.

You can also use Opus: "global.anthropic.claude-opus-4-5-20251101-v1:0"

qwen3-coder is having some func calling issue in OpenClaw.

Consider to opt in openai.gpt-oss-120b-1:0 with the same pricing rate.

Can you please provide whole config? can make it work(

I got error as below even exporting environment variable export AWS_BEARER_TOKEN_BEDROCK=xxxxxxxxx

error Failed to read config at /Users/9icloud/.openclaw/openclaw.json MissingEnvVarError: Missing env var "AWS_BEARER_TOKEN_BEDROCK" referenced at config path: models.providers.bedrock.apiKey
    at substituteString (file:///Users/9icloud/.nvm/versions/node/v24.13.0/lib/node_modules/openclaw/dist/config-BAN8QfW5.js:515:56)
    at substituteAny (file:///Users/9icloud/.nvm/versions/node/v24.13.0/lib/node_modules/openclaw/dist/config-BAN8QfW5.js:527:40)
    at substituteAny (file:///Users/9icloud/.nvm/versions/node/v24.13.0/lib/node_modules/openclaw/dist/config-BAN8QfW5.js:531:65)
    at substituteAny (file:///Users/9icloud/.nvm/versions/node/v24.13.0/lib/node_modules/openclaw/dist/config-BAN8QfW5.js:531:65)
    at substituteAny (file:///Users/9icloud/.nvm/versions/node/v24.13.0/lib/node_modules/openclaw/dist/config-BAN8QfW5.js:531:65)
    at substituteAny (file:///Users/9icloud/.nvm/versions/node/v24.13.0/lib/node_modules/openclaw/dist/config-BAN8QfW5.js:531:65)
    at resolveConfigEnvVars (file:///Users/9icloud/.nvm/versions/node/v24.13.0/lib/node_modules/openclaw/dist/config-BAN8QfW5.js:545:9)
    at Object.loadConfig (file:///Users/9icloud/.nvm/versions/node/v24.13.0/lib/node_modules/openclaw/dist/config-BAN8QfW5.js:4574:27)
    at loadConfig (file:///Users/9icloud/.nvm/versions/node/v24.13.0/lib/node_modules/openclaw/dist/config-BAN8QfW5.js:4825:20)
    at Object.stopChannel (file:///Users/9icloud/.nvm/versions/node/v24.13.0/lib/node_modules/openclaw/dist/gateway-cli-c_8Yf5s6.js:1792:15) {
  varName: 'AWS_BEARER_TOKEN_BEDROCK',
  configPath: 'models.providers.bedrock.apiKey'

One of solutions just use .env file based on official document
cat ~/.openclaw/.env

AWS_BEARER_TOKEN_BEDROCK={your_token}

Anyone can reference if you experience the issue
Thanks phaud's sharing

Hi everyone,

When I try to connect to AWS Bedrock via API, I keep getting the following error:

ThrottlingException
Too many tokens per day, please wait before trying again.

This is a brand-new AWS account, and I haven't actually been able to make a single successful request yet.
I originally assumed Bedrock would work on a pay-as-you-go basis, where usage is simply charged per token.

However, after searching online, I found many different explanations and suggested fixes, and I'm still not sure which configuration is actually correct or what needs to be enabled for a new account.

Has anyone experienced this with a new account?
Is there any quota, approval step, or default limit that needs to be adjusted before Bedrock API calls will work properly?

AWS new account will be rate limited on Bedrock
https://www.perplexity.ai/search/is-amazon-bedrock-rate-limitin-7f3rlnuaSI2lKIJ7TYFZ2A#0