Anthropic Launches Opus 5 With Near-Fable 5 Performance

As competition in generative artificial intelligence intensifies, the technology rivalry between unicorn startup Anthropic and OpenAI has become a focal point for the tech industry. Each upgrade to Anthropic’s flagship Claude series directly shapes the global cloud-computing ecosystem and the development of enterprise AI applications. It also serves as a key gauge of the future computing costs of commercializing AI.

Anthropic is reportedly preparing to unveil its new flagship model, Claude Opus 5, as early as mid-July 2026. The model is expected to offer lower costs while delivering performance close to Fable 5. A mysterious model code-named Honeycomb previously appeared briefly in the Cursor editor, where testing showed a context limit of 1 million tokens, positioning it to compete directly with OpenAI’s GPT-5.6.

Anthropic’s Claude Sonnet series is designed to balance speed, cost and reasoning capabilities for software development, AI agents and enterprise knowledge work. Sonnet 5 brings performance across several areas closer to the flagship Claude Opus 4.8, allowing developers to deploy long-running workflows and computer-use tasks without paying top-tier model rates. That improves the cost-effectiveness of enterprise generative AI adoption.

Anthropic launched Claude Sonnet 5 on June 30, 2026, with support for a 1 million-token context window and output of up to 128,000 tokens. Through August 31, promotional API pricing is $2 per million input tokens and $10 per million output tokens, 60% below Opus 4.8’s respective rates of $5 and $25. Prices will rise to $3 and $15 on September 1.

Anthropic has long competed in the generative AI market with its Claude model family, while its Mythos-class models were previously available only to governments and selected organizations. Claude Fable 5 brings capabilities in the same class to the public for the first time, marking a shift in advanced AI from closed deployments toward commercial use. The move is also drawing greater attention to software development, cybersecurity and biological-risk governance.

Anthropic unveiled Claude Fable 5 and Mythos 5 on June 10, claiming a performance improvement of more than 10% over the previous generation. Fable 5 is priced at twice the cost of Opus 4.8 and includes three layers of safety protections and model-distillation detection, while sensitive queries can be routed to other models. Mythos 5 remains restricted to governments and authorized organizations.

YouTuber 實測對比 Anthropic 旗下的 Fable 5 與 Claude Opus 4.8 兩款 AI 模型在 30 分鐘內架設網站與遊戲的表現。實測結果顯示，Fable 5 在執行速度、複雜邏輯架構處理及視覺精緻度上皆完勝 Opus 4.8，惟開發成本相對較高。

Anthropic's Claude Fable 5 was found to contain a rival-detection mechanism that downgraded the handling of development requests linked to other AI providers without notifying users. The issue raises concerns about fairness in model services, API predictability and transparency in safety policies, particularly for teams that rely on Claude to develop products.

Anthropic has publicly apologized for the mechanism that secretly weakened service for rival developers and pledged to end the opaque downgrading practice. If the system determines that a request poses a safety risk, it will instead explicitly fall back to Opus 4.8 or return a specific reason for refusal through the API. Reports did not disclose the date of the apology, when the changes would go live or any amount involved.

Anthropic positions its Claude Opus series as a high-end model for software development and agentic AI. For enterprises, the key question extends beyond benchmark scores to whether the model can reliably perform long-running tasks across multiple services. Opus 4.8 retains its predecessor’s pricing, lowering the barrier to upgrading, and targets engineering projects requiring extensive coordination, such as large-scale codebase migrations.

Anthropic released Claude Opus 4.8 on May 28, 2026, with its standard API priced at $5 per million input tokens and $25 per million output tokens. Official evaluations showed that the model was about one-quarter as likely as its predecessor to miss defects in code it had written. Claude Code also previewed Dynamic Workflows, which can run hundreds of subagents in parallel, and added an Effort control.

Anthropic launched its flagship Claude Opus 4.7 model in July 2026, targeting complex software engineering, image understanding and enterprise agent tasks. The upgrade also highlights how Anthropic divides responsibilities across its models: its public products balance utility and safety, while its most advanced cybersecurity capabilities are reserved for Claude Mythos under controlled testing.

Claude Opus 4.7 offers three times the visual resolution of its predecessor, along with stronger coding, instruction-following and long-running task performance. It also provides a 1 million-token context window at no extra charge. Anthropic said existing prompts may need adjustment because of changes to the model’s default behavior. Mythos, revealed at the same time, is available for testing only by selected cybersecurity organizations; no public release schedule or pricing has been announced.

Anthropic’s Claude Opus 4.6 and Sonnet 4.6 have expanded their context windows to 1 million tokens, enabling them to process large codebases, lengthy documents and multiple research materials in a single pass. The capability is critical to information integration, reasoning coherence and workflow efficiency as enterprises adopt generative AI, and has become a key competitive benchmark against OpenAI and Google.

Anthropic announced in 2026 that the 1 million-token context window for Opus 4.6 and Sonnet 4.6 had formally entered general availability, with existing pricing unchanged and no additional charge. Opus 4.6 scored 78.3% on the MRCR v2 long-context benchmark, significantly outperforming GPT-5.4 and Gemini 3.1 Pro.