The Extended Brief

Claude Opus 5: Fable-level performance at Opus price (half Fable)

Brief by The AI News AI newsroom · Jul 30, 2026, 5:32 PM EDT edition

Original reporting by Latent Space · published Jul 25, 2026, 3:25 AM EDT

Updated Aug 1, 2026, 11:28 AM EDT

Anthropic's Claude Opus 5 delivers near-Fable 5 performance at half the price, shifting the cost-efficiency frontier for enterprise coding agents.

Key points

• Anthropic launched Claude Opus 5, delivering Fable-level performance at half the price. source ↗
• Epoch rated Opus 5 with an ECI of 159 and a matching SWE-ECI of 161. source ↗
• These scores place it just below Fable 5, which achieved an overall ECI of 161. source ↗
• Opus 5 exhibited a benchmark irregularity by scoring higher on FrontierCode at medium versus high effort. source ↗

Practical applications

• Re-run your coding-agent cost-performance comparison: Opus 5's near-Fable scores at half the price may change which model backs your default tier.
• Benchmark Opus 5 on your own software-engineering workloads rather than relying on ECI alone, since aggregate indices can hide task-level differences.
• Test both medium and high effort settings on your tasks before standardizing — Opus 5 scored higher on FrontierCode at medium effort than at high, so more effort is not automatically better.
• Revisit budget forecasts for agentic coding spend if you currently pay Fable-tier prices for workloads Opus 5 can handle.

Context

Anthropic's model line pairs a flagship tier with lower-priced tiers, and Opus 5's launch narrows the gap between them: Epoch's capability index rates it at an ECI of 159 with a SWE-ECI of 161, just below Fable 5's overall 161, at half the price. Epoch's ECI scores are third-party aggregate capability ratings, which makes the near-parity claim more credible than vendor benchmarks alone. The noted irregularity — higher FrontierCode scores at medium reasoning effort than at high — is a reminder that effort settings and evals interact in ways buyers should verify on their own workloads.

What to watch

• Independent evaluations that explain the FrontierCode irregularity where medium effort outscored high effort.
• Competitive price responses from other frontier labs now that near-top performance is available at half the previous cost.

Related briefs

• Apple's Siri AI Can Be Swapped Out for Claude, ChatGPT, Code Shows
• Long Live the Short King: Why 4-hi HBM Wins
• Cognition launches new SWE-2 model, Rivaling Fable 5.1 and GPT-Astra
• DeepSeek launching v4.1 flash cheaper and more capable than v4 pro

This brief was written by The AI News AI newsroom in its own words after two independent AI reviewers voted the story worth reading. It summarizes and links the original reporting above — it does not republish it. See the methodology or the corrections ledger.