Vision

This guide describes how to work with images in Claude, including best practices, code examples, and limitations to keep in mind.



How to use vision

Use Claude's vision capabilities through:

• claude.ai. Upload an image like you would a file, or drag and drop an image directly into the chat window.
• The Console Workbench. A button to add images appears at the top right of every User message block.
• API request. See the examples in this guide.

Multiple images can be included in a single request, which Claude will analyze jointly when formulating its response. This can be helpful for comparing or contrasting images.



Before you upload



General limits

The maximal number of images per message or request is:

• 20 per message on claude.ai.
• 100 per request on the API, for models with a 200k-token context window.
• 600 per request on the API, for all other models.

The maximal dimensions per image are 8000x8000 px. If you submit more than 20 images in one API request, this limit is reduced to 2000x2000 px.

The maximal size per image is:

• 10 MB (base64-encoded) when using the Claude API directly.
• 5 MB (base64-encoded) on Amazon Bedrock and Vertex AI.
• 10 MB on claude.ai.



Evaluate image size

Claude views images in patches instead of pixels. Each patch is a 28×28 pixel block of the image, referred to as a visual token. An image therefore costs ⌈width / 28⌉ × ⌈height / 28⌉ visual tokens.

If Claude receives an image that is too large, it resizes it. The maximal native image resolution is:

• For Claude Fable 5 and Claude Mythos 5: 4784 tokens, and at most 2576 pixels on the long edge.
• For Claude Opus 4.8: 4784 tokens, and at most 2576 pixels on the long edge.
• For Claude Opus 4.7: 4784 tokens, and at most 2576 pixels on the long edge.
• For other models: 1568 tokens, and at most 1568 pixels on the long edge.

To minimize latency and to simplify coordinate-based workflows, you should prefer resizing images before uploading them.



Calculate image costs

Each image you include in a request to Claude counts toward your token usage. To calculate the approximate cost, multiply the image's visual token count (see Evaluate image size) by the per-token price of the model you're using.

Here are examples of tokenization and approximate costs for different image sizes within the API's size constraints based on Claude Sonnet 4.6 per-token price of $3 per million input tokens:

Note that the last three images exceed the native resolution and are downscaled before processing (to 1456x819 px, 1270x952 px, and 1456x819 px respectively), which caps their token cost. The 4K image costs no more than the 1920x1080 image because both downscale to the same size; the extra resolution is discarded.



High-resolution image support

Claude Opus 4.7 is the first Claude model with high-resolution image support; Claude Opus 4.8, Claude Fable 5, Claude Mythos 5, and later models also support it. The maximum image resolution is 2576 pixels on the long edge, up from 1568 px on prior models. This unlocks performance gains on vision-heavy workloads and is particularly valuable for computer use, screenshot understanding, and document analysis.

High-resolution support is automatic on Claude Opus 4.7 and later models and requires no beta header or client-side opt-in.

High-resolution images on Claude Opus 4.7, Claude Opus 4.8, Claude Fable 5, and Claude Mythos 5 can use up to approximately 3x more image tokens than on prior models (4784 versus 1568 tokens per image). If you don't need the additional fidelity, downsample images before sending to control token costs.

Here are the same image sizes tokenized for Claude Opus 4.7 and Claude Opus 4.8, based on their per-token price of $5 per million input tokens:

Only the last image exceeds the higher limits: the 4K image is downscaled to 2576x1449 px before processing. High-resolution support raises the resolution limits but does not remove them; images larger than 2576 px on the long edge (or 4784 visual tokens) are still downscaled.



Ensure image quality

When providing images to Claude, keep the following in mind for best results:

• Image format: Use a supported image format: JPEG, PNG, GIF, or WebP.
  Animations are unsupported, and only the first frame will be used.
• Image clarity: Ensure images are clear and not too blurry or pixelated.
• Text: If the image contains important text, make sure it's legible and not too small. Avoid cropping out key visual context just to enlarge the text.
• Resizing: Take into account that your image might be resized if it is too large (see above); this might for example make text less legible. Consider pre-resizing your images, cropping them, or both.
• Image compression: Compressing images before sending them, using a lossy format such as JPEG or WebP (lossy mode), can reduce latency by reducing the size of requests. However, this can introduce artifacts that are detrimental to model performance, especially when multiple compression passes are applied. For example, heavy JPEG compression can make text difficult to read. Confirm your compression settings are appropriate for the task by inspecting the actual images sent to the API.



Working with coordinates and bounding boxes

Claude can locate and label regions of an image (for example, returning bounding boxes for tables, form fields, chart elements, or UI components).

Coordinates follow the standard image convention: the origin (0, 0) is the top-left corner of the image, with x increasing to the right and y increasing downward. The coordinates Claude returns are pixel positions in the image Claude sees: your image after Claude resizes it to fit the model's native resolution (see How Claude resizes and pads images). To get coordinates you can use directly, either pre-resize your image so the coordinates map one-to-one onto the image you have (see Resize your image before uploading), or rescale the coordinates Claude returns (see Rescale coordinates when you cannot pre-resize).



How Claude resizes and pads images

Claude finds the largest aspect-preserving size that satisfies both of the model's image limits:

1. Edge limit: neither side exceeds the maximum edge length (1568 px for most models, 2576 px for Claude Opus 4.7 and later models).
2. Visual token limit: the image's token cost ⌈width / 28⌉ × ⌈height / 28⌉ does not exceed the model's visual token budget (1568 tokens for most models, 4784 for Claude Opus 4.7 and later models).

For most photos and screenshots the edge limit is what triggers a resize. For portrait documents the visual token limit usually triggers first, and overlooking it is the most common cause of misaligned coordinates. For example, an A4 page scanned at 130 DPI is 1075×1520 pixels: both sides are under 1568 px, but it costs 39 × 55 = 2145 visual tokens, so Claude resizes it to 924×1307.

Claude then pads every image, whether or not it was resized, up to the next multiple of 28 pixels on the bottom and right edges (924×1307 becomes 924×1316 in the example). The padding contains no content: Claude perceives the padded image, but the page content only ever occupies the un-padded resized region. Always normalize or rescale by the resized dimensions, not the padded dimensions; dividing by the padded dimensions scales every coordinate by a small amount.



Resize your image before uploading

The most reliable approach is to resize your image yourself before uploading, so the image you have is exactly the image Claude sees and the coordinates Claude returns need no conversion.

The following reference implementation computes the exact size Claude resizes an image to:

import math


def count_image_tokens(width: int, height: int) -> int:
    """Visual tokens consumed by an image: one token per 28x28 pixel patch."""
    return math.ceil(width / 28) * math.ceil(height / 28)


def resized_size(
    width: int,
    height: int,
    max_edge: int = 1568,
    max_tokens: int = 1568,
) -> tuple[int, int]:
    """The size Claude resizes an image to before padding.

    Defaults are for most models. For Claude Opus 4.7 and later models, use
    max_edge=2576 and max_tokens=4784. Returns (width, height). Images that
    already fit within the limits are returned unchanged.
    """

    def fits(w: int, h: int) -> bool:
        return (
            math.ceil(w / 28) * 28 <= max_edge
            and math.ceil(h / 28) * 28 <= max_edge
            and count_image_tokens(w, h) <= max_tokens
        )

    if fits(width, height):
        return (width, height)
    if height > width:
        resized_h, resized_w = resized_size(height, width, max_edge, max_tokens)
        return (resized_w, resized_h)

    # Binary search along the long edge for the largest aspect-preserving
    # size that fits.
    aspect_ratio = width / height
    lo, hi = 1, width  # lo always fits; hi never fits
    while lo + 1 < hi:
        mid = (lo + hi) // 2
        if fits(mid, max(round(mid / aspect_ratio), 1)):
            lo = mid
        else:
            hi = mid
    return (lo, max(round(lo / aspect_ratio), 1))


# The A4 example from "How Claude resizes and pads images":
print(resized_size(1075, 1520))  # (924, 1307)

1. Resize the image to the dimensions returned by resized_size. If the image already fits within the model's limits, resized_size returns its dimensions unchanged and no resize is needed.
2. Send the resized image to the API. Don't pad it yourself; Claude handles padding, and padding doesn't shift the coordinate origin.
3. In your prompt, ask explicitly for pixel coordinates. For example: "Return the bounding box of each table as [x1, y1, x2, y2] in pixel coordinates."
4. Use the returned coordinates directly against the image you sent. If you need normalized coordinates, divide by the dimensions of the image you sent, not by the original image's dimensions and not by the padded dimensions.



Rescale coordinates when you cannot pre-resize

If you cannot pre-resize (for example, when the image comes from an upstream system you can't modify), use resized_size from Resize your image before uploading to recover the dimensions Claude saw, then map the coordinates Claude returns into normalized coordinates or back onto your original image. This approach requires knowing the pixel dimensions of the image you uploaded, so it does not apply to PDF uploads.

def to_relative_coordinates(
    x: float,
    y: float,
    original_width: int,
    original_height: int,
    max_edge: int = 1568,
    max_tokens: int = 1568,
) -> tuple[float, float]:
    """Map a pixel coordinate returned by Claude to relative coordinates in [0, 1].

    Pass the dimensions of the image you uploaded. For Claude Opus 4.7 and
    later models, use max_edge=2576 and max_tokens=4784.
    """
    resized_w, resized_h = resized_size(
        original_width, original_height, max_edge, max_tokens
    )
    return (x / resized_w, y / resized_h)


# To express the coordinate in your original image's pixel space, multiply the
# relative coordinate by your original dimensions:
# (rel_x * original_width, rel_y * original_height)

Padding is applied only to the bottom and right edges, so the origin doesn't shift and a per-axis linear rescale is sufficient.



Prompt examples

Many of the prompting techniques that work well for text-based interactions with Claude can also be applied to image-based prompts.

These examples demonstrate best practice prompt structures involving images.



About the prompt examples

The following examples demonstrate how to use Claude's vision capabilities using various programming languages and approaches. You can provide images to Claude in three ways:

1. As a base64-encoded image in image content blocks
2. As a URL reference to an image hosted online
3. Using the Files API (upload once, use multiple times)

The base64 example prompts use these variables:

import base64
import httpx

# For base64-encoded images
image1_url = "https://upload.wikimedia.org/wikipedia/commons/a/a7/Camponotus_flavomarginatus_ant.jpg"
image1_media_type = "image/jpeg"
image1_data = base64.standard_b64encode(httpx.get(image1_url).content).decode("utf-8")

image2_url = "https://upload.wikimedia.org/wikipedia/commons/b/b5/Iridescent.green.sweat.bee1.jpg"
image2_media_type = "image/jpeg"
image2_data = base64.standard_b64encode(httpx.get(image2_url).content).decode("utf-8")

# For URL-based images, you can use the URLs directly in your requests

Below are examples of how to include images in a Messages API request using base64-encoded images and URL references:



Base64-encoded image example

image1_data = "iVBORw0KGgoAAAANSUhEUgAAAAEAAAABCAIAAACQd1PeAAAADElEQVR4nGP4z8AAAAMBAQDJ/pLvAAAAAElFTkSuQmCC"
image1_media_type = "image/png"

client = anthropic.Anthropic()
message = client.messages.create(
    model="claude-opus-4-8",
    max_tokens=1024,
    messages=[
        {
            "role": "user",
            "content": [
                {
                    "type": "image",
                    "source": {
                        "type": "base64",
                        "media_type": image1_media_type,
                        "data": image1_data,
                    },
                },
                {"type": "text", "text": "Describe this image."},
            ],
        }
    ],
)
print(message)



URL-based image example

client = anthropic.Anthropic()
message = client.messages.create(
    model="claude-opus-4-8",
    max_tokens=1024,
    messages=[
        {
            "role": "user",
            "content": [
                {
                    "type": "image",
                    "source": {
                        "type": "url",
                        "url": "https://upload.wikimedia.org/wikipedia/commons/a/a7/Camponotus_flavomarginatus_ant.jpg",
                    },
                },
                {"type": "text", "text": "Describe this image."},
            ],
        }
    ],
)
print(message)



Files API image example

For images you'll use repeatedly or when you want to avoid encoding overhead, use the Files API. Upload the image once, then reference the returned file_id in subsequent messages instead of resending base64 data.

client = anthropic.Anthropic()

# Upload the image file
with open("image.jpg", "rb") as f:
    file_upload = client.beta.files.upload(file=("image.jpg", f, "image/jpeg"))

# Use the uploaded file in a message
message = client.beta.messages.create(
    model="claude-opus-4-8",
    max_tokens=1024,
    betas=["files-api-2025-04-14"],
    messages=[
        {
            "role": "user",
            "content": [
                {
                    "type": "image",
                    "source": {"type": "file", "file_id": file_upload.id},
                },
                {"type": "text", "text": "Describe this image."},
            ],
        }
    ],
)

print(message.content)

See Messages API examples for more example code and parameter details.



Limitations

While Claude's image understanding capabilities are cutting-edge, there are some limitations to be aware of:

• People identification: Claude cannot be used to name people in images and refuses to do so.
• Accuracy: Claude may hallucinate or make mistakes when interpreting low-quality, rotated, or very small images under 200 pixels.
• Spatial reasoning: Claude's coordinate and localization outputs are approximate. Follow the guidance in Working with coordinates and bounding boxes and verify outputs before relying on them.
• Counting: Claude can give approximate counts of objects in an image but may not always be precisely accurate, especially with large numbers of small objects.
• AI generated images: Claude does not know if an image is AI-generated and may be incorrect if asked. Do not rely on it to detect fake or synthetic images.
• Inappropriate content: Claude does not process inappropriate or explicit images that violate the Acceptable Use Policy.
• Healthcare applications: While Claude can analyze general medical images, it is not designed to interpret complex diagnostic scans such as CTs or MRIs. Claude's outputs should not be considered a substitute for professional medical advice or diagnosis.

Always carefully review and verify Claude's image interpretations, especially for high-stakes use cases. Do not use Claude for tasks requiring perfect precision or sensitive image analysis without human oversight.



FAQ



Dive deeper into vision

Ready to start building with images using Claude? Here are a few helpful resources:

• Multimodal cookbook: This cookbook has tips on getting started with images and best practice techniques to ensure the highest quality performance with images. See how you can effectively prompt Claude with images to carry out tasks such as interpreting and analyzing charts or extracting content from forms.
• API reference: Documentation for the Messages API, including example API calls involving images.

If you have any other questions, reach out to the support team. You can also join the developer community to connect with other creators and get help from Anthropic experts.