Status (July 24, 2026): Claude Opus 5 has not been officially released. Anthropic's model documentation still lists Claude Opus 4.8 (May 28, 2026) as the current Opus flagship. No Opus 5 announcement, model card, API identifier, or pricing page exists. A model codenamed "Honeycomb" surfaced briefly in Cursor's model picker on July 8 and was pulled within hours.
On July 23, deployment trackers reported launch preparations, and a Cursor error dialog on July 24 referenced the string claude-opus-5-thinking-high. Neither amounts to an official release. The best models available right now are Fable 5 for the hardest tasks, Opus 4.8 for complex work, and Sonnet 5 for everyday building.
What Is Claude Opus 5?
Claude Opus 5 is the expected next-generation model in Anthropic's Opus tier, the company's premium line for complex reasoning, coding, and autonomous agent work. The current Opus flagship is Claude Opus 4.8, released on May 28, 2026.
Anthropic's model lineup currently breaks into distinct tiers. At the top sits the Mythos class, which includes Claude Fable 5 (launched June 9, 2026) and the restricted Claude Mythos 5. Below that is the Opus tier, where Opus 4.8 holds the line. Then comes Sonnet 5 (June 30, 2026), Anthropic's high-value workhorse, followed by Haiku 4.5 for lightweight, high-volume tasks.
If Opus 5 ships, it would slot between the Mythos tier and the current Opus 4.8, likely offering stronger performance than 4.8 at a more accessible price point than Fable 5's $10/$50 per million tokens.
What Is Confirmed vs. What Is Rumor
Confirmed:
Opus 4.8 remains the newest Opus model, available at $5/$25 per million input/output tokens. Fable 5 is the most capable model Anthropic has released publicly. Sonnet 5 launched June 30, 2026 and serves as the default model on free and Pro Claude plans. Anthropic's news page and model documentation contain no Opus 5 entry.
Reported but unverified:
On July 8-9, a model called "Claude Honeycomb EAP" appeared inside Cursor's model picker and was pulled within hours. It was a real listing in a shipping product, seen by multiple people and reported by The New Stack before it disappeared.
The listing showed a 1M-token context window, an "extra high effort" reasoning mode (a step above the "high" default on Opus 4.8), per-turn controls with safety fallbacks that route to Opus 4.8, and placement alongside other frontier models in Cursor's picker.
Two things stood out. The fallback chain stepping down to Opus 4.8 implies Honeycomb sits above 4.8 in capability. Fallbacks conventionally step down, not up. And Cursor has been a reliable pre-launch venue before. A "claude-sonnet-5" identifier surfaced in Google Vertex AI logs before Sonnet 5 launched, following the same pattern.
The listing did not include pricing, benchmarks, an official name, or a release date. "Honeycomb" is a codename. It could ship as Opus 5, join the Mythos tier, arrive under a different name, or never ship publicly at all.
Then came July 23. Deployment trackers reported Opus 5 preparations and described the model as "rolling out across providers." Some users on X claimed they were served improved responses under the Opus 4.8 label. On July 24, a Cursor error dialog surfaced showing the string claude-opus-5-thinking-high, the first appearance of the literal "claude-opus-5" name in a shipping product. None of these carry an official Anthropic source, and every rumored date so far has passed without an official launch.
What Claude Models Are Available Right Now
Claude Fable 5 is Anthropic's most capable publicly available model. It excels at the hardest autonomous tasks: long-running code migrations, complex multi-step research, and anything where getting it right on the first pass saves significant time and money. It costs $10/$50 per million input/output tokens and requires 30-day data retention. Think of this as the specialist you bring in for the toughest jobs.
Claude Opus 4.8 is the reliable workhorse for complex work. It handles multi-step coding, agent workflows, and enterprise tasks well, at $5/$25 per million tokens. It supports zero data retention, making it the choice for organizations with strict compliance requirements. For most professional work, Opus 4.8 is where quality and cost find a good balance.
Claude Sonnet 5 is the value pick that has been closing the gap fast. At $3/$15 per million tokens (with introductory pricing of $2/$10 through August 31, 2026), it delivers performance close to Opus 4.8 on many tasks. It is the default model on Claude's free and Pro plans and the best starting point for most everyday work.
Claude Haiku 4.5 is the speed and budget option at $1/$5 per million tokens. It handles simple classification, extraction, and high-volume tasks where cost matters more than depth.
For most builders, the practical strategy is straightforward: start with Sonnet 5 for everyday work, step up to Opus 4.8 when accuracy on hard tasks matters, and reserve Fable 5 for the work where failure costs more than the token premium.
What This Means If You Build Without Code
A stronger Opus model means better code generation, fewer iteration cycles, and more reliable handling of complex app logic. That matters whether you are building a booking system, a CRM, or a client portal. But it is a background improvement, not something that requires you to change your tools or learn a new API.
The current model lineup is the strongest it has ever been. Sonnet 5 alone outperforms what was available at the Opus tier just six months ago. If you have been waiting for "the next model" to start a project, the current options are more than capable.
On Emergent, Sonnet 5, Fable 5, Opus 4.8, and all other major models are already available to build with. When Opus 5 ships, it will be too.
Stay tuned to Emergent News for more on AI tools, launches, and what they mean for builders.
Emergent turns your idea into a full-stack web or mobile app, no coding required.
No coding required- Web & mobile apps
- Deploys instantly
on Emergent today
