Claude Opus 4.8: Anthropic's Flagship Agentic Coding Model
Claude Opus 4.8 is Anthropic's flagship model, released May 28, 2026, and positioned as the best agentic coding model available. It plans before it edits, holds 1 million tokens of context at a flat rate, and runs long autonomous tasks without losing the thread. It is built for builders and operators who push hard on big codebases, tool-using agents, and long-context analysis.
What Claude Opus 4.8 is best for
Opus 4.8 is the top of Anthropic's lineup. It is the strongest agentic coding model in its class, and it plans before editing rather than guessing its way through changes. That planning step is what makes it hold up on long autonomous runs, where weaker models drift.
- Agentic coding across large, messy codebases
- Long-context analysis over big document sets and repos
- Production assistants and RAG pipelines
- Tool-using agents that chain many steps
For hard agentic coding or analysis, compare Opus 4.8 with Sonnet 5 on your own workload. Sonnet 5 is now the default on Claude Free and Pro and is available in Claude Code and the API, so it should be the current baseline before you pay for or route to another model.
Context window
Opus 4.8 carries 1 million tokens of context at a flat rate, with no long-context surcharge. That means you can load entire codebases or large document sets and reason over them in one pass. Sonnet 4.6 also offers 1M tokens; Haiku 4.5 tops out at 200K.
Pricing
Opus 4.8 is the premium tier. API pricing is $5 per million input tokens and $25 per million output tokens. Fast Mode doubles that to $10 in and $50 out. Batch processing runs 50% cheaper, and prompt caching cuts cached input cost by up to 90%. For comparison, Sonnet 4.6 is $3 in / $15 out and Haiku 4.5 is $1 in / $5 out.
How it compares
- Opus 4.8: flagship, best for agentic coding and heavy long-context work, premium price
- Sonnet 5: the current default on Claude Free and Pro, available in Claude Code and the API, with introductory API pricing of $2/$10 through August 31, 2026 and $3/$15 after
- Haiku 4.5: fastest and cheapest, 200K context, for high-volume light tasks
Should you use it
Choose Opus 4.8 when the job is genuinely hard: long autonomous coding runs, deep reasoning over a full codebase, or agents that need to plan and use tools reliably. If you can get the result on Sonnet 4.6, use Sonnet and save the spend. Note the gaps: no native image generation (Opus is text and code focused), voice exists in the mobile apps but is not a strength, and the premium price adds up at scale. Web search and image input are both supported, and Anthropic created the Model Context Protocol with first-class MCP support, with Claude Code as the dedicated CLI.
Specs at a Glance
| Spec | Claude Opus 4.8 |
|---|---|
| Released | May 28, 2026 |
| Context window | 1M tokens (flat rate, no surcharge) |
| Input price (per M tokens) | $5 ($10 Fast Mode) |
| Output price (per M tokens) | $25 ($50 Fast Mode) |
| Best for | Agentic coding, long-context analysis, tool-using agents |
| Image generation | No (text and code focused) |
| Web search | Yes |
Get the Claude playbook in your inbox.
One weekly email for Claude and Claude Code users. Real workflows, no hype. Subscribe and we send you The Claude Power-User Cheatsheet.