Claude Opus 5: What's New, Pricing, and How It Compares to Fable 5

Anthropic launched Claude Opus 5 on July 24, 2026, at half the price of Fable 5. Here is what changed, what it costs, and where it wins and loses.

By Quantara Editorial Team · AI Tools · Published Jul 25, 2026 · 8 min read

Anthropic released Claude Opus 5 on July 24, 2026, as claude-opus-5 across the API, Claude.ai, Claude Code, and Claude Cowork. It is now the default model on Claude Max and the strongest option available on Claude Pro. Unlike Fable 5, which arrived and was suspended within days over export-control concerns, Opus 5 shipped as a straightforward general-availability release.

Claude Opus 5 Pricing

Anthropic priced Opus 5 at $5 per million input tokens and $25 per million output tokens, matching the old Opus 4.8 rate card and landing at exactly half of Fable 5's $10/$50 pricing. A "fast mode" is also available at roughly double the standard price for about 2.5x faster responses on latency-sensitive workloads.

Benchmark Results

Anthropic's own Frontier-Bench results describe Opus 5 as more than doubling Opus 4.8's performance at a lower cost per task. On CursorBench 3.2, a coding benchmark, Opus 5 lands within roughly 0.5% of Fable 5's peak score while costing about half as much per task. On ARC-AGI 3, Anthropic reports Opus 5 scoring roughly three times higher than the next-best model, and on OSWorld 2.0, a computer-use benchmark, Opus 5 beats Fable 5's best recorded result at about a third of the cost.

What Changed Under the Hood

Opus 5 supports mid-conversation tool changes without invalidating the prompt cache, which matters for long agent sessions that add or swap tools as a task evolves. Anthropic also carried over Fable 5's automatic safety-classifier routing, which can redirect certain high-risk requests to a different model rather than answering directly.

How Does Safety Compare to Fable 5 and Mythos 5?

Anthropic reports Opus 5's misaligned-behavior score at 2.3, the lowest of its recent model releases, but says the model remains "substantially behind Mythos 5" specifically on cyber-exploitation resistance. In practice this means Opus 5 ships without the extra safeguard tier and 30-day retention requirement that applied to Fable 5 and Mythos 5, which is part of why it launched without an export-control review.

Should You Switch From Fable 5?

For coding and computer-use workloads, Opus 5's combination of near-parity benchmark scores and half the price makes it the more economical default now that Fable 5 has returned to the market as a premium option. Teams running the heaviest reasoning workloads that specifically need Fable 5's or Mythos 5's extra safeguards, or its slightly higher peak coding scores, may still prefer to keep that model in the rotation for those specific tasks.

Build Prompts for Opus 5

Use Verbito's AI Prompt Generator to structure agentic and coding prompts that specify goals, constraints, tool access, and verification steps so they carry over cleanly whether you route the task to Opus 5 or another model.

Sources

• Anthropic: Introducing Claude Opus 5
• 9to5Mac: Anthropic upgrades Claude with new Opus 5 model
• ExplainX: Claude Opus 5 launch, July 2026