Claude 3 Opus
Claude 3 Opus is the highest-capability tier in the Claude 3 family from Anthropic. It targets complex tasks with strong recall across a context window of 200K tokens and a twofold accuracy improvement over Claude 2.1 on open-ended questions.
import { streamText } from 'ai'
const result = streamText({ model: 'anthropic/claude-3-opus', prompt: 'Why is the sky blue?'})About Claude 3 Opus
Claude 3 Opus launched on March 1, 2023 as the highest-capability model in the Claude 3 family, alongside Sonnet and Haiku. Haiku prioritized throughput. Sonnet balanced capability with efficiency. Opus targeted tasks where intelligence was the primary constraint.
Accuracy on difficult factual questions defined the release. Compared to Claude 2.1, Opus showed a twofold improvement in correct answers on complex open-ended queries while reducing hallucinated responses. It produced more right answers and fewer confidently wrong ones. Anthropic also reported near-perfect recall (exceeding 99% accuracy) on the Needle In A Haystack evaluation for long context. In some cases, it identified that test needles appeared artificially inserted, a sign of meta-reasoning rather than pure retrieval.
Opus launched with vision capabilities matching other comparable models at the time, processing photos, charts, graphs, and technical diagrams. All Claude 3 models improved at following complex, multi-step instructions and producing structured output in formats like JSON compared to earlier generations. Anthropic noted Opus adhered better to brand voice guidelines, a consideration for enterprises building customer-facing products.
The Claude 3 family positioned Opus for research and development workflows, complex task automation, strategy analysis, and scientific reasoning. Anthropic indicated that inputs exceeding 1 million tokens were technically possible and available to select customers, even though the standard context window is 200K tokens.