10 replies

Trouble with connecting to a openrouter model

✅Solved

I'm trying to connect an openrouter model openrouter/tngtech/deepseek-r1t2-chimera:free, but I am getting the following error:

Agent failed before reply: Unknown model: openrouter/tngtech/deepseek-r1t2-chimera:free.
Logs: openclaw logs --follow
I tried tngtech/deepseek-r1t2-chimera:free, have the money on the account. I have successfully connected to openrouter/deepseek/deepseek-chat. What is the issue?

I'm also getting this error with openrouter/openrouter/pony-alpha

"openclaw models list" puts out the following:

Model Input Ctx Local Auth Tags
openrouter/openrouter/pony-alpha text 195k no yes default,configured
openrouter/deepseek/deepseek-chat text 160k no yes fallback#1,configured
tngtech/deepseek-r1t2-chimera:free - - - - configured,missing
openrouter/tngtech/deepseek-r1t2-chimer... - - - - configured,missing
openrouter/pony-alpha - - - - configured,missing

Solution

Yep — you can “teach” OpenClaw a missing OpenRouter model by adding a custom provider under

models.providers

models.providers

.

Important caveat: if you name the provider

openrouter

openrouter

, it becomes a full replacement for the built‑in OpenRouter catalog (so you’d have to list all OpenRouter models you want). So the safe approach is: create a new provider name, e.g.

openrouter-manual

openrouter-manual

.

Add something like this to

~/.openclaw/openclaw.json

~/.openclaw/openclaw.json

(JSON5):

{
  models: {
    mode: "merge",
    providers: {
      "openrouter-manual": {
        baseUrl: "https://openrouter.ai/api/v1",
        api: "openai-completions",

        // Option A (recommended): read from env var OPENROUTER_API_KEY

{
  models: {
    mode: "merge",
    providers: {
      "openrouter-manual": {
        baseUrl: "https://openrouter.ai/api/v1",
        api: "openai-completions",

        // Option A (recommended): read from env var OPENROUTER_API_KEY

Jump to solution

FotC�

Friends of the Crustacean 🦞🤝•3d ago•

10 replies

disk

Trouble with connecting to a openrouter model

✅Solved

I'm trying to connect an openrouter model openrouter/tngtech/deepseek-r1t2-chimera:free, but I am getting the following error:

Solution

Yep — you can “teach” OpenClaw a missing OpenRouter model by adding a custom provider under

models.providers

models.providers

.

Important caveat: if you name the provider

openrouter

openrouter

, it becomes a full replacement for the built‑in OpenRouter catalog (so you’d have to list all OpenRouter models you want). So the safe approach is: create a new provider name, e.g.

openrouter-manual

openrouter-manual

.

Add something like this to

~/.openclaw/openclaw.json

~/.openclaw/openclaw.json

(JSON5):

{
  models: {
    mode: "merge",
    providers: {
      "openrouter-manual": {
        baseUrl: "https://openrouter.ai/api/v1",
        api: "openai-completions",

        // Option A (recommended): read from env var OPENROUTER_API_KEY

{
  models: {
    mode: "merge",
    providers: {
      "openrouter-manual": {
        baseUrl: "https://openrouter.ai/api/v1",
        api: "openai-completions",

        // Option A (recommended): read from env var OPENROUTER_API_KEY

Jump to solution

Trouble with connecting to a openrouter model

Trouble with connecting to a openrouter model

Similar Threads

Similar Threads

Similar Threads