Claude Opus 4.8 Is Now Available in Puter.js

Puter.js now supports Claude Opus 4.8, Anthropic's most capable model—built for honest, reliable agentic work, sharper coding judgment, and stronger multimodal reasoning.

What is Claude Opus 4.8?

Claude Opus 4.8 is Anthropic's latest flagship model, released on May 28, 2026. It builds on Opus 4.7 with major gains in honesty, tool-calling efficiency, and long-session reliability, while keeping the same 1M context window and $5/$25 per-million-token pricing.

Key highlights:

• Honesty and reliability gains — roughly 4x less likely to let code flaws pass unremarked, with better judgment about when to ask clarifying questions and flag mistakes
• More efficient tool calling — completes agentic workflows in fewer steps with better context retention across long sessions
• Stronger multimodal reasoning — improved analysis over PDFs, diagrams, and charts, with higher citation precision
• 1 million token context window — process entire codebases, lengthy documents, and complex datasets in a single prompt
• 128,000 output tokens — generate long-form analyses, refactors, and detailed reports in one response
• Dynamic workflows — the model can plan, branch, and adapt its own multi-step process inside a single task

What's New Compared to Opus 4.7?

Opus 4.8 focuses on making agentic work more trustworthy and efficient:

• Leader on Terminal-Bench 2.1 — sets a new state of the art for terminal-driven agentic tasks
• 82.3% on OSWorld-Verified — significantly improved computer-use accuracy in real desktop environments
• 84% on Online-Mind2Web — strong gains on browser-agent tasks
• CursorBench leader across effort levels — exceeds prior Opus models at every effort setting for coding work
• First model to break 10% on the Legal Agent Benchmark all-pass standard — meaningful progress on long-horizon legal reasoning
• Only model completing all Super-Agent cases end-to-end — finishes the full agentic benchmark suite without dropping tasks

Opus 4.8 also introduces fast mode pricing at $10/$50 per million tokens—about 3x cheaper than previous Opus fast configurations.

Examples

Agentic Coding with Stronger Judgment

puter.ai.chat([
  { role: "user", content: largeCodebase + "\n\nRefactor the auth module to use the new session API. Flag anything ambiguous before changing it." }
], { model: 'claude-opus-4-8' });

Multimodal Document Analysis

puter.ai.chat([
  { role: "user", content: [
    { type: "file", puter_path: "~/Q1-report.pdf" },
    { type: "text", text: "Extract the key financial metrics, cite the page numbers, and flag any inconsistencies between the charts and the narrative." }
  ]}
], { model: 'claude-opus-4-8' });

Streaming

const response = await puter.ai.chat(
  "Design a distributed task scheduler with fault tolerance and exactly-once execution guarantees",
  { model: 'claude-opus-4-8', stream: true }
);

for await (const part of response) {
  puter.print(part?.text);
}

Get Started Now

Just add one library to your project:

// npm install @heyputer/puter.js
import { puter } from '@heyputer/puter.js';

Or add one script tag to your HTML:

<script src="https://js.puter.com/v2/"></script>

No API keys and no infrastructure setup. Start building with Claude Opus 4.8 immediately.

Learn more:

• Free, Unlimited Claude API
• Puter.js AI Documentation