inferrs

inferrs can serve local models behind an OpenAI-compatible /v1 API. Velaclaw works with inferrs through the generic openai-completions path. inferrs is currently best treated as a custom self-hosted OpenAI-compatible backend, not a dedicated Velaclaw provider plugin.

Getting started

Start inferrs with a model

inferrs serve google/gemma-4-E2B-it \
  --host 127.0.0.1 \
  --port 8080 \
  --device metal

Verify the server is reachable

curl http://127.0.0.1:8080/health
curl http://127.0.0.1:8080/v1/models

Add an Velaclaw provider entry

Add an explicit provider entry and point your default model at it. See the full config example below.

Full config example

This example uses Gemma 4 on a local inferrs server.

{
  agents: {
    defaults: {
      model: { primary: "inferrs/google/gemma-4-E2B-it" },
      models: {
        "inferrs/google/gemma-4-E2B-it": {
          alias: "Gemma 4 (inferrs)",
        },
      },
    },
  },
  models: {
    mode: "merge",
    providers: {
      inferrs: {
        baseUrl: "http://127.0.0.1:8080/v1",
        apiKey: "inferrs-local",
        api: "openai-completions",
        models: [
          {
            id: "google/gemma-4-E2B-it",
            name: "Gemma 4 E2B (inferrs)",
            reasoning: false,
            input: ["text"],
            cost: { input: 0, output: 0, cacheRead: 0, cacheWrite: 0 },
            contextWindow: 131072,
            maxTokens: 4096,
            compat: {
              requiresStringContent: true,
            },
          },
        ],
      },
    },
  },
}

Advanced

Why requiresStringContent matters

Some inferrs Chat Completions routes accept only string messages[].content, not structured content-part arrays.

If Velaclaw runs fail with an error like:

messages[1].content: invalid type: sequence, expected a string

set compat.requiresStringContent: true in your model entry.

compat: {
  requiresStringContent: true
}

Velaclaw will flatten pure text content parts into plain strings before sending the request.

Gemma and tool-schema caveat

Some current inferrs + Gemma combinations accept small direct /v1/chat/completions requests but still fail on full Velaclaw agent-runtime turns.If that happens, try this first:

compat: {
  requiresStringContent: true,
  supportsTools: false
}

That disables Velaclaw’s tool schema surface for the model and can reduce prompt pressure on stricter local backends.If tiny direct requests still work but normal Velaclaw agent turns continue to crash inside inferrs, the remaining issue is usually upstream model/server behavior rather than Velaclaw’s transport layer.

Manual smoke test

Once configured, test both layers:

curl http://127.0.0.1:8080/v1/chat/completions \
  -H 'content-type: application/json' \
  -d '{"model":"google/gemma-4-E2B-it","messages":[{"role":"user","content":"What is 2 + 2?"}],"stream":false}'

velaclaw infer model run \
  --model inferrs/google/gemma-4-E2B-it \
  --prompt "What is 2 + 2? Reply with one short sentence." \
  --json

If the first command works but the second fails, check the troubleshooting section below.

Proxy-style behavior

inferrs is treated as a proxy-style OpenAI-compatible /v1 backend, not a native OpenAI endpoint.

Native OpenAI-only request shaping does not apply here
No service_tier, no Responses store, no prompt-cache hints, and no OpenAI reasoning-compat payload shaping
Hidden Velaclaw attribution headers (originator, version, User-Agent) are not injected on custom inferrs base URLs

Troubleshooting

curl /v1/models fails

inferrs is not running, not reachable, or not bound to the expected host/port. Make sure the server is started and listening on the address you configured.

messages[].content expected a string

Set compat.requiresStringContent: true in the model entry. See the requiresStringContent section above for details.

Direct /v1/chat/completions calls pass but velaclaw infer model run fails

Try setting compat.supportsTools: false to disable the tool schema surface. See the Gemma tool-schema caveat above.

inferrs still crashes on larger agent turns

If Velaclaw no longer gets schema errors but inferrs still crashes on larger agent turns, treat it as an upstream inferrs or model limitation. Reduce prompt pressure or switch to a different local backend or model.

For general help, see Troubleshooting and FAQ.

Local models

Running Velaclaw against local model servers.

Gateway troubleshooting

Debugging local OpenAI-compatible backends that pass probes but fail agent runs.

Model providers

Overview of all providers, model refs, and failover behavior.

Overview

Concepts and configuration

Providers

inferrs

inferrs

Getting started

Full config example

Advanced

Troubleshooting

See also

Local models

Gateway troubleshooting

Model providers

​inferrs

​Getting started

​Full config example

​Advanced

​Troubleshooting

​See also

Local models

Gateway troubleshooting

Model providers

inferrs

Getting started

Full config example

Advanced

Troubleshooting

See also