Editing article

Title

Summary

Content

Google Cloud recently announced that <a href="https://www.anthropic.com/news/claude-3-family">Anthropic’s Claude 3 Models</a> will be available on Google Cloud (Sonnet and Haiku for now), and Opus will be added in coming weeks.<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*GTJap7mGMYHggm-I-NUxnQ.png" /></figure><h3>What is Claude 3?</h3>Claude 3 comes with 3 state-of-the-art models: Opus, Sonnet and Haiku.Opus: Excels in complex tasks understanding new situations with impressive human-like ability. It pushes the boundaries of what AI can do.Sonnet: Balances performance and cost, well-suited for businesses needing fast and reliable performance.Haiku: Super fast and small, giving lightning quick answers to questions. This lets you create AI that feels like talking to a real person.<h3>Comparison of the Claude 3 models with GPT and Gemini Models.</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*DhSMUYbSBL-rnZT0" /></figure><h3>Cost Comparison between 3 Models:</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*XhCvOEiuCnbnyUlr" /></figure><h3>How to get started?</h3><ol><li>Go to Model Garden tab in Vertex AI for <a href="https://console.cloud.google.com/vertex-ai/publishers/anthropic/model-garden/claude-3-sonnet">Sonnet</a> or <a href="https://console.cloud.google.com/vertex-ai/publishers/anthropic/model-garden/claude-3-haiku">Haiku</a></li><li>Click on Enable, Fill the basic details about your Organization or just about your self.</li><li>Once Step 2 is done, wait for 2–3 minutes to get the model enabled.</li><li>You can now view the code for interacting with Claude 3 models. It has a lot of examples like Text input, Image input, Streaming responses, Calling via API, Calling via SDK.</li></ol><h3>Steps Interact with Claude 3 using SDK</h3><pre>!pip3 install anthropic[vertex]</pre><pre>MODEL = &quot;claude-3-sonnet@20240229&quot; #for Haiku claude-3-haiku@20240307 REGION = &quot;us-central1&quot; PROJECT_ID = &quot;[your-project-id]&quot; import vertexai import json vertexai.init(project=PROJECT_ID, location=REGION)</pre><h3>1. Text Input</h3><pre>from anthropic import AnthropicVertex client = AnthropicVertex(region=REGION, project_id=PROJECT_ID) message = client.messages.create( max_tokens=1024, messages=[ { &quot;role&quot;: &quot;user&quot;, &quot;content&quot;: &quot;Send me a recipe for Pizza.&quot;, } ], model=MODEL, ) data = json.loads(message.model_dump_json(indent=2))[&quot;content&quot;][0] print(data[&quot;text&quot;])</pre>Query: Send me a recipe for Pizza.Response:<figure><img alt="" src="https://cdn-images-1.medium.com/max/838/1*fqH4D2F-_EMrsXMk3MSvVw.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*f7E7aEVYYScPmxMKxEhbYQ.png" /></figure><h3>2. Single Image Input</h3>Image Used:<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*lnsWMpIz61oidGEiqgktZg.jpeg" /><figcaption>Image 1</figcaption></figure><pre>import base64 import httpx from anthropic import AnthropicVertex client = AnthropicVertex(region=REGION, project_id=PROJECT_ID) image1_url = &quot;https://cache.getarchive.net/Prod/thumb/cdn12/L3Bob3RvLzIwMTYvMTIvMzEvdHJhZmZpYy1qYW0tdHJhZmZpYy1pbmRpYS10cmFuc3BvcnRhdGlvbi10cmFmZmljLTFlNDJiZi0xMDI0LmpwZw%3D%3D/1280/720/jpg&quot; image1_media_type = &quot;image/jpeg&quot; image1_data = base64.b64encode(httpx.get(image1_url).content).decode(&quot;utf-8&quot;) message = client.messages.create( max_tokens=1024, messages=[ { &quot;role&quot;: &quot;user&quot;, &quot;content&quot;: [ { &quot;type&quot;: &quot;image&quot;, &quot;source&quot;: { &quot;type&quot;: &quot;base64&quot;, &quot;media_type&quot;: image1_media_type, &quot;data&quot;: image1_data, }, }, {&quot;type&quot;: &quot;text&quot;, &quot;text&quot;: &quot;Describe the image and get the location and weather.&quot;}, ], } ], model=MODEL, ) data = json.loads(message.model_dump_json(indent=2))[&quot;content&quot;][0] print(data[&quot;text&quot;])</pre>Query: Describe the image and get the location and weather.Response:<figure><img alt="" src="https://cdn-images-1.medium.com/max/881/1*TkPjrrUhBYFkrKUlgH2y_Q.png" /></figure><h3>3. Multi Image Input:</h3>Images Used:<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*E2Nc-I6HA6Qyvu1G" /><figcaption>Image 1</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*YPtg3Jiw3HEx8I88" /><figcaption>Image 2</figcaption></figure><pre>import base64 import httpx from anthropic import AnthropicVertex client = AnthropicVertex(region=REGION, project_id=PROJECT_ID) image1_url = &quot;https://parkplus.io/_next/image?url=https%3A%2F%2Fstrapi-file-uploads.s3.ap-south-1.amazonaws.com%2Fopen_top_cars_2da902c4b3.jpg&amp;w=1920&amp;q=75&quot; image2_url = &quot;https://images.pexels.com/photos/3817871/pexels-photo-3817871.jpeg?auto=compress&amp;cs=tinysrgb&amp;w=1260&amp;h=750&amp;dpr=2&quot; image1_media_type = &quot;image/jpeg&quot; image1_data = base64.b64encode(httpx.get(image1_url).content).decode(&quot;utf-8&quot;) image2_data = base64.b64encode(httpx.get(image2_url).content).decode(&quot;utf-8&quot;) message = client.messages.create( max_tokens=1024, messages=[ { &quot;role&quot;: &quot;user&quot;, &quot;content&quot;: [ { &quot;type&quot;: &quot;image&quot;, &quot;source&quot;: { &quot;type&quot;: &quot;base64&quot;, &quot;media_type&quot;: image1_media_type, &quot;data&quot;: image1_data, }, }, { &quot;type&quot;: &quot;image&quot;, &quot;source&quot;: { &quot;type&quot;: &quot;base64&quot;, &quot;media_type&quot;: image1_media_type, &quot;data&quot;: image2_data, }, }, {&quot;type&quot;: &quot;text&quot;, &quot;text&quot;: &quot;What are the similarities and differences between these two images&quot;}, ], } ], model=MODEL, ) data = json.loads(message.model_dump_json(indent=2))[&quot;content&quot;][0] print(data[&quot;text&quot;])</pre>Query: What are the similarities and differences between these two imagesResponse:<figure><img alt="" src="https://cdn-images-1.medium.com/max/885/1*ZiFv1QIZDyJcQbCSXyVTaQ.png" /></figure><h3>Conclusion:</h3>Claude 3 has Capabilities like:<ol><li>Performing complex cognitive tasks.</li><li>Transcribing and analyzing static images.</li><li>Code generation</li><li>Translating between various languages in real-time.</li><li>Offers models with combination of speed and performance depending upon the use case.</li><li>It is Secure, Capable and Reliable.</li></ol><h3>If you liked this post, please Clap for it. Follow me if you want to read more such posts!</h3><h3>Twitter: <a href="https://twitter.com/IVaibhavMalpani">https://twitter.com/IVaibhavMalpani</a> LinkedIn: <a href="https://www.linkedin.com/in/ivaibhavmalpani/">https://www.linkedin.com/in/ivaibhavmalpani/</a></h3><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=20c65b308f01" width="1" height="1" alt=""><hr><a href="https://medium.com/google-cloud/claude-3-on-google-cloud-20c65b308f01">Getting Started with Claude 3 on Google Cloud</a> was originally published in <a href="https://medium.com/google-cloud">Google Cloud - Community</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.

Author

Link

Published date

Image url

Feed url

Guid

Hidden blurb

--- !ruby/object:Feedjira::Parser::RSSEntry
title: Getting Started with Claude 3 on Google Cloud
url: https://medium.com/google-cloud/claude-3-on-google-cloud-20c65b308f01?source=rss----e52cf94d98af---4
author: Vaibhav Malpani
categories:
- anthropic-claude
- multimodal
- generative-ai
- gemini
- google-cloud-platform
published: 2024-04-01 04:11:36.000000000 Z
entry_id: !ruby/object:Feedjira::Parser::GloballyUniqueIdentifier
 is_perma_link: 'false'
 guid: https://medium.com/p/20c65b308f01
carlessian_info:
 news_filer_version: 2
 newspaper: Google Cloud - Medium
 macro_region: Blogs
rss_fields:
- title
- url
- author
- categories
- published
- entry_id
- content
content: 'Google Cloud recently announced that <a href="https://www.anthropic.com/news/claude-3-family">Anthropic’s
 Claude 3 Models</a> will be available on Google Cloud (Sonnet and Haiku for now),
 and Opus will be added in coming weeks.<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*GTJap7mGMYHggm-I-NUxnQ.png"
 /></figure><h3>What is Claude 3?</h3>Claude 3 comes with 3 state-of-the-art models:
 Opus, Sonnet and Haiku.Opus: Excels in complex tasks understanding
 new situations with impressive human-like ability. It pushes the boundaries of what
 AI can do.Sonnet: Balances performance and cost, well-suited
 for businesses needing fast and reliable performance.Haiku: Super
 fast and small, giving lightning quick answers to questions. This lets you create
 AI that feels like talking to a real person.<h3>Comparison of the Claude
 3 models with GPT and Gemini Models.</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*DhSMUYbSBL-rnZT0"
 /></figure><h3>Cost Comparison between 3 Models:</h3><figure><img
 alt="" src="https://cdn-images-1.medium.com/max/1024/0*XhCvOEiuCnbnyUlr" /></figure><h3>How
 to get started?</h3><ol><li>Go to Model Garden tab in Vertex AI for <a
 href="https://console.cloud.google.com/vertex-ai/publishers/anthropic/model-garden/claude-3-sonnet">Sonnet</a>
 or <a href="https://console.cloud.google.com/vertex-ai/publishers/anthropic/model-garden/claude-3-haiku">Haiku</a></li><li>Click
 on Enable, Fill the basic details about your Organization or just about your self.</li><li>Once
 Step 2 is done, wait for 2–3 minutes to get the model enabled.</li><li>You can now
 view the code for interacting with Claude 3 models. It has a lot of examples like
 Text input, Image input, Streaming responses, Calling via API, Calling via SDK.</li></ol><h3>Steps
 Interact with Claude 3 using SDK</h3><pre>!pip3 install anthropic[vertex]</pre><pre>MODEL
 = &quot;claude-3-sonnet@20240229&quot; #for Haiku claude-3-haiku@20240307 REGION
 = &quot;us-central1&quot; PROJECT_ID = &quot;[your-project-id]&quot; import
 vertexai import json vertexai.init(project=PROJECT_ID, location=REGION)</pre><h3>1.
 Text Input</h3><pre>from anthropic import AnthropicVertex client = AnthropicVertex(region=REGION,
 project_id=PROJECT_ID) message = client.messages.create( max_tokens=1024, messages=[ { &quot;role&quot;:
 &quot;user&quot;, &quot;content&quot;: &quot;Send me a recipe for
 Pizza.&quot;, } ], model=MODEL, ) data = json.loads(message.model_dump_json(indent=2))[&quot;content&quot;][0] print(data[&quot;text&quot;])</pre>Query:
 Send me a recipe for Pizza.Response:<figure><img
 alt="" src="https://cdn-images-1.medium.com/max/838/1*fqH4D2F-_EMrsXMk3MSvVw.png"
 /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*f7E7aEVYYScPmxMKxEhbYQ.png"
 /></figure><h3>2. Single Image Input</h3>Image Used:<figure><img
 alt="" src="https://cdn-images-1.medium.com/max/1024/1*lnsWMpIz61oidGEiqgktZg.jpeg"
 /><figcaption>Image 1</figcaption></figure><pre>import base64 import httpx from
 anthropic import AnthropicVertex client = AnthropicVertex(region=REGION,
 project_id=PROJECT_ID) image1_url = &quot;https://cache.getarchive.net/Prod/thumb/cdn12/L3Bob3RvLzIwMTYvMTIvMzEvdHJhZmZpYy1qYW0tdHJhZmZpYy1pbmRpYS10cmFuc3BvcnRhdGlvbi10cmFmZmljLTFlNDJiZi0xMDI0LmpwZw%3D%3D/1280/720/jpg&quot; image1_media_type
 = &quot;image/jpeg&quot; image1_data = base64.b64encode(httpx.get(image1_url).content).decode(&quot;utf-8&quot;) message
 = client.messages.create( max_tokens=1024, messages=[ { &quot;role&quot;:
 &quot;user&quot;, &quot;content&quot;: [ { &quot;type&quot;:
 &quot;image&quot;, &quot;source&quot;: { &quot;type&quot;:
 &quot;base64&quot;, &quot;media_type&quot;: image1_media_type, &quot;data&quot;:
 image1_data, }, }, {&quot;type&quot;:
 &quot;text&quot;, &quot;text&quot;: &quot;Describe the image and get the location
 and weather.&quot;}, ], } ], model=MODEL, ) data
 = json.loads(message.model_dump_json(indent=2))[&quot;content&quot;][0] print(data[&quot;text&quot;])</pre>Query:
 Describe the image and get the location and weather.Response:<figure><img
 alt="" src="https://cdn-images-1.medium.com/max/881/1*TkPjrrUhBYFkrKUlgH2y_Q.png"
 /></figure><h3>3. Multi Image Input:</h3>Images Used:<figure><img
 alt="" src="https://cdn-images-1.medium.com/max/1024/0*E2Nc-I6HA6Qyvu1G" /><figcaption>Image
 1</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*YPtg3Jiw3HEx8I88"
 /><figcaption>Image 2</figcaption></figure><pre>import base64 import httpx from
 anthropic import AnthropicVertex client = AnthropicVertex(region=REGION,
 project_id=PROJECT_ID) image1_url = &quot;https://parkplus.io/_next/image?url=https%3A%2F%2Fstrapi-file-uploads.s3.ap-south-1.amazonaws.com%2Fopen_top_cars_2da902c4b3.jpg&amp;w=1920&amp;q=75&quot; image2_url
 = &quot;https://images.pexels.com/photos/3817871/pexels-photo-3817871.jpeg?auto=compress&amp;cs=tinysrgb&amp;w=1260&amp;h=750&amp;dpr=2&quot; image1_media_type
 = &quot;image/jpeg&quot; image1_data = base64.b64encode(httpx.get(image1_url).content).decode(&quot;utf-8&quot;) image2_data
 = base64.b64encode(httpx.get(image2_url).content).decode(&quot;utf-8&quot;) message
 = client.messages.create( max_tokens=1024, messages=[ { &quot;role&quot;:
 &quot;user&quot;, &quot;content&quot;: [ { &quot;type&quot;:
 &quot;image&quot;, &quot;source&quot;: { &quot;type&quot;:
 &quot;base64&quot;, &quot;media_type&quot;: image1_media_type, &quot;data&quot;:
 image1_data, }, }, { &quot;type&quot;:
 &quot;image&quot;, &quot;source&quot;: { &quot;type&quot;:
 &quot;base64&quot;, &quot;media_type&quot;: image1_media_type, &quot;data&quot;:
 image2_data, }, }, {&quot;type&quot;:
 &quot;text&quot;, &quot;text&quot;: &quot;What are the similarities and differences
 between these two images&quot;}, ], } ], model=MODEL, ) data
 = json.loads(message.model_dump_json(indent=2))[&quot;content&quot;][0] print(data[&quot;text&quot;])</pre>Query:
 What are the similarities and differences between these two imagesResponse:<figure><img
 alt="" src="https://cdn-images-1.medium.com/max/885/1*ZiFv1QIZDyJcQbCSXyVTaQ.png"
 /></figure><h3>Conclusion:</h3>Claude 3 has Capabilities like:<ol><li>Performing
 complex cognitive tasks.</li><li>Transcribing and analyzing static images.</li><li>Code
 generation</li><li>Translating between various languages in real-time.</li><li>Offers
 models with combination of speed and performance depending upon the use case.</li><li>It
 is Secure, Capable and Reliable.</li></ol><h3>If you liked this post, please Clap
 for it. Follow me if you want to read more such posts!</h3><h3>Twitter: <a href="https://twitter.com/IVaibhavMalpani">https://twitter.com/IVaibhavMalpani</a> LinkedIn:
 <a href="https://www.linkedin.com/in/ivaibhavmalpani/">https://www.linkedin.com/in/ivaibhavmalpani/</a></h3><img
 src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=20c65b308f01"
 width="1" height="1" alt=""><hr><a href="https://medium.com/google-cloud/claude-3-on-google-cloud-20c65b308f01">Getting
 Started with Claude 3 on Google Cloud</a> was originally published in <a href="https://medium.com/google-cloud">Google
 Cloud - Community</a> on Medium, where people are continuing the conversation by
 highlighting and responding to this story.'

Language

Active

Ricc internal notes

Ricc source

Show this article Back to articles