Claude Opus 5: Near-Fable Intelligence at Half the Price

Anthropic released Claude Opus 5 on July 24, 2026 — the fourth Claude 5 model in under two months, following Mythos 5, Fable 5, and Sonnet 5 in June. It's a thoughtful and proactive model that comes close to the frontier intelligence of Claude Fable 5 at half the price.

What's new in Opus 5

• Outperforms Fable 5 on coding and knowledge work — Opus 5 outperforms Fable 5 on coding and knowledge work evaluations like Frontier-Bench and GDPval-AA, scoring higher on 8 of 13 benchmarks despite costing about 50% less.
• Effort dial — users can toggle how much effort — low, medium, or high — the model expends completing a task. This enables users to balance between cost and capability. At lower effort levels, Opus 5 uses fewer tokens while preserving much of its performance.
• Self-verification — the model verifies its own work and recovers from errors without intervention.
• Mid-task model switching — users can switch which model they are using mid-task, which can be used to keep AI costs in check.
• Safety improvements — Opus 5 is "the most aligned Opus model and is the least susceptible to being tricked into misuse."
• Life sciences — Opus 5 outperformed Opus 4.8 across all internal life sciences benchmarks, proving particularly adept at automated protein research.

Pricing

Opus 5 delivers significantly more capability than Opus 4.8 at identical pricing — making it an automatic upgrade for any team already on Opus 4.8.

The Anthropic model hierarchy in July 2026

Opus 5 vs Fable 5: when to use which

The effort dial and token costs

The effort dial is the most significant cost-management feature Opus 5 introduces. At low effort, Opus 5 generates shorter reasoning traces and fewer output tokens — approaching Sonnet-level cost for straightforward tasks while preserving Opus-level quality for tasks that need it.

This is a meaningful shift for production applications: instead of routing simple queries to a cheaper model manually, you can use a single Opus 5 endpoint and dial down effort for low-complexity requests automatically.

Using Opus 5 with Tokoscope

Opus 5 uses the model string claude-opus-5-20260724. Tokoscope tracks it the same way as any Claude model:

from tokoscope import wrap
import anthropic

client = wrap(anthropic.Anthropic(), api_key="ts_live_...")

message = client.messages.create(
    model="claude-opus-5-20260724",
    max_tokens=1024,
    messages=[{"role": "user", "content": "Hello"}]
)