Anthropic Lists Claude Opus 4.8 With 1M-Token Context and $5/$25 API Pricing

Anthropic has listed Claude Opus 4.8 as a generally available model on Claude and the Claude Platform, according to the company’s product page and API documentation.

What Anthropic says about Claude Opus 4.8

Anthropic’s Claude Opus 4.8 product page describes the model as the company’s most capable generally available model for coding, agentic tasks, and knowledge-work capabilities. The same Anthropic page says Claude Opus 4.8 is available through Claude and the Claude Platform.

Anthropic’s API documentation lists the model under the API ID claude-opus-4-8. The documentation also states that Claude Opus 4.8 supports a 1 million-token context window, includes adaptive thinking, and has a reliable knowledge cutoff of January 2026.

For developers and businesses comparing model costs, Anthropic’s product page and API documentation both list Claude Opus 4.8 at $5 per million input tokens and $25 per million output tokens. Those prices place it in the same published pricing structure across Anthropic’s marketing page and technical documentation.

Benchmark reporting

All Things How published an article on May 31, 2026 titled “Claude Opus 4.8 Benchmarks: Scores, Rankings, and Pricing (May 2026).” The article reports benchmark rankings for Claude Opus 4.8 and also cites the model’s $5 per million input-token and $25 per million output-token pricing, along with its 1 million-token context window.

Because benchmark results can depend heavily on test selection, evaluation methodology, prompting, and whether results are reported by a vendor or a third party, the All Things How benchmark rankings should be read alongside Anthropic’s own model documentation rather than as a complete measure of real-world performance. The provided Anthropic sources confirm the model’s availability, API identifier, pricing, context window, adaptive thinking feature, and stated knowledge cutoff; they do not, in the supplied excerpts, provide the detailed benchmark table referenced by All Things How.

Why the 1 million-token context window matters

A 1 million-token context window can allow a model to process much larger bodies of text in a single request than shorter-context systems, depending on implementation and product limits. In practice, that can be relevant for tasks such as reviewing long codebases, analyzing extensive documents, or working across large collections of internal material. Anthropic’s API documentation is the source for the 1 million-token context figure for Claude Opus 4.8.

The presence of adaptive thinking in Anthropic’s API documentation also indicates that the model supports Anthropic’s reasoning-related configuration for supported use cases. The supplied documentation excerpt does not specify performance gains from adaptive thinking, so the feature should be understood as a documented capability rather than a guaranteed improvement for every task.

Pricing and availability

Anthropic’s published pricing for Claude Opus 4.8 is straightforward: $5 per million input tokens and $25 per million output tokens. The company’s product page says the model is available on Claude and the Claude Platform, while the API documentation provides the model ID developers would use: claude-opus-4-8.

For organizations evaluating Claude Opus 4.8, the key confirmed details are its general availability through Anthropic’s services, its 1 million-token context window, its January 2026 reliable knowledge cutoff, and its published API pricing. Benchmark rankings reported by All Things How may be useful as a starting point, but buyers and developers should compare those results with their own workloads and constraints before drawing conclusions.