The model does not support network images

Error message

The model does not support network images: {network_image_urls:?}

What it means

The Bedrock request builder collects image URLs found in user messages that are network URLs (http/https) rather than base64 data. Since Bedrock's InvokeModel API requires images to be passed inline as base64 content, network-hosted image URLs cannot be forwarded, so the builder fails before any request is sent.

Solutions

Download the image, base64-encode it, and pass it as inline data (ImageMediaType with decoded data) instead of a URL
Preprocess messages to convert network image URLs to base64 before calling the client
Use a provider that supports URL images (e.g. OpenAI) if inline images are not feasible
Strip image content from messages when targeting Bedrock models

Example fix

// before
UserContent::Image(ImageUrl::Url("https://example.com/cat.png".into()))
// after
let bytes = reqwest::get(url).await?.bytes().await?;
UserContent::Image(ImageUrl::Decode(String::from("png"), base64::encode(bytes)))

Defensive patterns

Strategy: validation

Validate before calling

let network_imgs: Vec<_> = messages.iter()
    .flat_map(|m| m.images())
    .filter(|img| img.url().map_or(false, |u| u.starts_with("http")))
    .collect();
if !network_imgs.is_empty() {
    // download + base64-encode before sending
}

Type guard

fn is_network_image(img: &Image) -> bool {
    matches!(img, Image::Url(u) if u.starts_with("http"))
}

Prevention

Always convert remote image URLs to base64 inline data for Bedrock
Build a preprocessing step in the message pipeline
Add a unit test that rejects http image URLs for Bedrock targets
Prefer providers that accept URL images when URLs are required

When it happens

Trigger: build_chat_completions_body (via chat_completions_builder) finds a non-empty network_image_urls list — i.e. a UserContent message contains an image whose source is a remote URL instead of embedded base64 data.

Common situations: Passing image URLs scraped from the web directly in chat messages, forgetting to download and base64-encode images before calling Bedrock-backed models, or porting code written for OpenAI (which accepts URLs) to Bedrock.

Understand the failure class

Background: UnsupportedOperationException and "is not supported" errors: when a library deliberately refuses a call — this error's family across 30 libraries.

Related errors

AI-assisted analysis of sigoden/aichat@82976d349a (2026-09-09). Data as JSON: /api/errors/055f3435b91b8fe5. Report an issue: GitHub.

Appendix: source

Thrown at src/client/bedrock.rs:431

                        }));
                    }
                    vec![
                        json!({
                            "role": "assistant",
                            "content": assistant_parts,
                        }),
                        json!({
                            "role": "user",
                            "content": user_parts,
                        }),
                    ]
                }
            }
        })
        .collect();

    if !network_image_urls.is_empty() {
        bail!(
            "The model does not support network images: {:?}",
            network_image_urls
        );
    }

    let mut body = json!({
        "inferenceConfig": {},
        "messages": messages,
    });
    if let Some(v) = system_message {
        body["system"] = json!([
            {
                "text": v,
            }
        ])
    }

    if let Some(v) = model.max_tokens_param() {

View on GitHub (pinned to 82976d349a)