Claude Haiku 4.5
Anthropic's fast, cost-efficient model with strong reasoning, coding, and tool use at low latency.
Claude Haiku 4.5 is a cloud text model from Anthropic. It is multimodal: alongside a text prompt it accepts images, then replies with generated text. Its context window handles up to 200K input tokens and up to 64K output tokens, and it supports adjustable reasoning effort for harder problems. Generation can be tuned with system instructions and temperature. It runs through Replicate and Runware using your own API key, from $1.00 per million input tokens.
- Pricing
- $1.00 / 1M in · $5.00 / 1M out
- Inputs
- Text, Images
- Context window
- 200K in · 64K out
- Reasoning
- Adjustable effort
- Controls
- System prompt, Temperature
Examples
Generated with Claude Haiku 4.5 via Runware. The same three prompts run against every text model in the catalog, shown verbatim — so the only thing that changes between two models’ answers is the model.
Explain what an API is to someone who has never written code. Do not use a restaurant or waiter analogy. Plain prose only — no headings, bullet points, or markdown formatting. Maximum 50 words.
An API is a tool that lets different software programs talk to each other and share information. Think of it as a messenger that takes your request, delivers it to another program, and brings back the answer you need. APIs make it possible for apps to work together smoothly.
A two-person startup is deciding whether to build their own user authentication or use a hosted provider. Give them a clear recommendation, then the single strongest argument against your own recommendation. Plain prose only — no headings, bullet points, or markdown formatting. Maximum 150 words.
You should use a hosted authentication provider. With two people, you cannot afford the engineering time to build authentication securely. Hosted providers handle password hashing, token management, breach response, and compliance standards that would take you months to implement correctly. They're also cheaper than paying developers to maintain custom auth code over time. The strongest argument against this is that authentication is core to your product's security model. By delegating it, you depend on a third party's infrastructure, pricing decisions, and policy changes. If your provider raises rates significantly or experiences an outage, you have limited recourse. You also lose direct control over user data and authentication flows, which matters for compliance-sensitive applications. Building internally means you own your security destiny, even if it's more work initially.
Write a briefing for a non-technical executive explaining why their company's spending on AI is hard to predict, and what they can do about it. Cover at least three distinct causes. Plain prose only — no headings, bullet points, or markdown formatting. Write exactly 205 words.
Your company's AI spending is difficult to predict due to three fundamental factors that distinguish artificial intelligence from traditional technology investments. First, AI projects have highly variable timelines. While conventional software follows established development patterns, AI requires experimentation and iteration. Models may underperform unexpectedly, requiring substantial additional investment to improve accuracy. You cannot know upfront how many iterations you'll need before achieving acceptable results. Second, computational costs fluctuate based on factors beyond your control. As your models scale or process larger datasets, infrastructure expenses can spike dramatically. Cloud providers adjust pricing, and newer hardware becomes available, creating constant pressure to upgrade systems. Unlike software licensing with fixed contracts, these costs remain dynamic and difficult to forecast. Third, talent retention significantly impacts budgets. AI specialists command premium salaries in a highly competitive market, and losing key personnel means retraining costs and project delays. Unexpected departures force accelerated hiring at premium rates, inflating expenses unpredictably. To manage these uncertainties, adopt a staged funding approach rather than committing to fixed annual budgets. Build contingency reserves of 20-30 percent into AI initiatives. Establish clear performance milestones that trigger go-or-no-go decisions before committing additional resources. Finally, partner with external consultants to validate project feasibility early, reducing the risk of expensive course corrections later.
Anthropic
Anthropic builds the Claude family of large language models, focused on safety, helpfulness, and honesty.
www.anthropic.com ↗