As of July 2026, Anthropic offers four current model tiers designed for different workloads. Understanding their capabilities and trade-offs is essential for cost-effective production systems.
| Model | Best For | Context | Max Output | Pricing (in/out MTok) |
|---|---|---|---|---|
| Claude Fable 5 (claude-fable-5) | Long-running agents, highest capability | 1M | 128K | $10 / $50 |
| Claude Opus 4.8 (claude-opus-4-8) | Complex agentic coding, enterprise work | 1M | 128K | $5 / $25 |
| Claude Sonnet 5 (claude-sonnet-5) | Best speed/intelligence balance | 1M | 128K | $3 / $15 (intro $2 / $10 until Aug 31, 2026) |
| Claude Haiku 4.5 (claude-haiku-4-5) | High volume, low latency, classification | 200K | 64K | $1 / $5 |
Generally available since June 9, 2026, Claude Fable 5 is Anthropic's most capable widely released model, built for long-running agents. Adaptive thinking is always on and cannot be disabled. Its sibling Claude Mythos 5 shares Fable 5's specs but is invitation-only for defensive cybersecurity work under Project Glasswing.
Released May 28, 2026, Opus 4.8 introduces Adaptive Thinking — the model dynamically decides when deeper reasoning is required based on task complexity, with the effort parameter defaulting to high on all surfaces. Substantially improved vision capabilities support higher image resolution for more accurate analysis of charts, dense documents, and complex UI screens. Note: Opus 4.8 (and Fable 5) use the tokenizer introduced with Opus 4.7, which produces roughly 30% more tokens than pre-4.7 models on the same text; re-benchmark your cost estimates when migrating.
The 1M token context window is now standard (no beta header) on Fable 5, Opus 4.8/4.7/4.6, Sonnet 5, and Sonnet 4.6 — Haiku 4.5 remains at 200K. This allows analysis of entire codebases, multi-hundred-page legal documents, or massive datasets in a single request. Max output is 128K tokens on the Messages API (64K on Haiku), extendable to 300K on the Batch API via the output-300k-2026-03-24 beta header.