The release of the Anthropic film Claude Haiku 4.5, a latency-optimized “small” Model that Delivers Claude Sonnet 4, which has similar performance levels, is a good example of a coder. While running Double the performance at only a third the price. Anthropic API’s and catalogs of partners are already offering the model. Amazon Bedrock The following are some examples of how to get started: Google Cloud Vertex AI. Prices are Input of $1/MTok and output of $5/MTok. Haiku 4.5 is positioned by Anthropic as a replacement for Haiku 3.5 The following are some examples of how to get started: Sonnet 4 in cost-sensitive, interactive workloads.
Positioning and line-up
Haiku 4.5 is a real-time assistant, automations for customer support, and pair programming where latency budgets are tight and throughput rules. It surpasses Sonnet 4 on “computer use” tasks—the GUI/browser manipulation underpinning products like Claude for Chrome—and is described as materially improving responsiveness in Claude Code Anthropic is a great choice for rapid prototyping and multi-agent project. The Anthropic website makes it clear that Sonnet 4/5 remains as the Frontier Model “the best coding model in the world,” Haiku 4.5 is available now Near-Frontier performance at a lower cost. This is a recommended pattern. Sonnet 4.5, for planning in multiple steps The following are some examples of how to get started: Haiku workers execute in parallel.
Prices, availability and identification
Developers can start calling the model from day one.claude-haiku-4-5Anthropic also states availability on the API. Anthropic states that it is also available on Amazon Bedrock The following are some examples of how to get started: Vertex AIThe model catalogs will update over time with updated region and IDs, but in its launch post the company confirmed cloud support. The launch post confirms cloud availability. The API Price Haiku 4.5 has been released. $1/MTok (input) The following are some examples of how to get started: $5/MTok (output)There is a list of quick-caching at 1 MTok for $2.25 The following are some examples of how to get started: Read for $0.10 per MTok.
Benchmarks
The Anthropic suite includes a summary of results for both the standard and agentic suites, as well as methodology information to help qualify those numbers.
- SWE-bench VerifiedSimple scaffolding with two tools: (bash, edit files), 73.3% Over average Test 50, no test-time compute, Think budget 128K, default sampling. Adds a small prompt to encourage extensive tool usage and first writing of tests.
- Terminal-BenchTerminus-2 Agent, Average over 11 Runs (6 Without Thinking, 5 With Budget 32k).
- OSWorld-Verified: Maximum 100 stepsThe average of 4 runs using a Think Budget: $128K The following are some examples of how to get started: 2K per-step configuration.
- AIME / MMLUAverages across multiple runs with default sampling Think budgets 128K.
This post highlights Coding parity with Sonnet 4, The following are some examples of how to get started: Computer use gains Sonnet 4 on these scaffolds. Before generalizing, users should try to replicate their orchestration and tool stacks.
What you need to know
- Haiku delivers code performance comparable with Sonnet-4, at less than a third the cost. It also doubles the speed.
- It is faster than Sonnet 4 in computer tasks.
- Haiku 4.5 can be used to parallelise execution and plan multiple-step plans using Sonnet.
- The pricing is $5/$1 per million tokens of input/output. It’s available through Claude API and Google Cloud Vertex AI.
- In Anthropic’s tests, this release was released under ASL-2 and had a measured misalignment percentage lower than Opus 4.1 or Sonnet 4.5.
Anthropic’s positioning of Claude Haiku 4.5 was strategically sound. By delivering Claude Sonnet 5 has similar coding performances to Claude Sonnet 4. As you can see, the following is a list of some helpful websites. Three-quarters of the cost The following are some examples of how to get started: More than double the speed. Sonnet 4: Computer use is superior, the company gives devs a clean planner–executor split—Sonnet 4.5 Multi-step planning, and pool of Haiku 4.5 workers for parallel execution—without forcing architectural changes (“drop-in replacement” You can find out more about this by clicking here. API, Amazon Bedrock, Vertex AI). The ASL-2 A release accompanied by a documentation lower misalignment rate Then, Sonnet 4.5 The following are some examples of how to get started: The Opus 4.1Reduces friction during enterprise rollouts, where the deployment equation is dominated by safety barriers and budget envelopes.
Take a look at the Technical details, system card, model page” documentation . Please feel free to browse our GitHub Page for Tutorials, Codes and Notebooks. Also, feel free to follow us on Twitter Join our Facebook group! 100k+ ML SubReddit Subscribe now our Newsletter. Wait! Are you using Telegram? now you can join us on telegram as well.
Asif Razzaq, CEO of Marktechpost Media Inc. is a visionary engineer and entrepreneur who is dedicated to harnessing Artificial Intelligence’s potential for the social good. Marktechpost was his most recent venture. This platform, which focuses on machine learning and deep-learning news, is both technical and understandable to a broad audience. Over 2 million views per month are a testament to the platform’s popularity.
