Claude Opus 5: Release Date, What We Know & Full Claude Model Guide

Last updated: July 23, 2026

Claude Opus 5 does not exist. Anthropic has not released, announced, or teased a model with that name. If you searched for “Claude Opus 5” expecting the next Opus after 4.8, here is what actually happened: Anthropic introduced a new model tier above Opus called Mythos-class, and the first generally available Mythos-class model is Claude Fable 5, released on June 9, 2026.

The latest Opus model remains Claude Opus 4.8, released on May 28, 2026.

Quick answer: There is no Claude Opus 5. The model many people expected to be “Opus 5” is actually Claude Fable 5 — a Mythos-class model that sits above Opus in Anthropic’s lineup. Fable 5 costs $10/$50 per million input/output tokens, has a 1M-token context window, 128K max output, and scores significantly above Opus 4.8 on agentic coding benchmarks. Opus 4.8 remains the best Opus-class model at $5/$25 per MTok.

Why there is no Claude Opus 5

The Claude model naming changed in 2026. Instead of incrementing Opus to version 5, Anthropic created a new capability tier above Opus called Mythos-class. The first publicly available Mythos-class model is Claude Fable 5.

This is not a simple rename. The tier structure now looks like this:

Anthropic’s reasoning is straightforward: the capability jump from Opus 4.8 to the next generation was large enough to warrant a new tier name rather than a version bump. Fable 5 is not “Opus but better” — it is a different class of model with different safeguards, different data-retention requirements, and a different price point.

The naming is confusing, and that is exactly why “Claude Opus 5” is one of the most common search queries in the Claude ecosystem. People expect the next Opus to be called Opus 5. Instead, Anthropic leapfrogged the name entirely.

For the deep-dive on Fable 5 itself — release date, safeguards, API details, GitHub Copilot access — read the full Claude Fable 5 guide.

The full Claude model lineup in July 2026

Here is every current Claude model with the specs that matter for developers and teams.

All models support text and image input, text output, vision, tool use, and multilingual capabilities. They are available through the Claude API, Amazon Bedrock, Claude Platform on AWS, Google Cloud, and Microsoft Foundry.

Legacy models still available

If you are still on Opus 4.7 or earlier, the upgrade path is clear: Opus 4.8 for Opus-class work, Fable 5 for the hardest tasks, Sonnet 5 for cost-sensitive production.

Claude Fable 5: the model people mean when they search “Opus 5”

When most people search for “Claude Opus 5,” they want the next-generation model above Opus 4.8. That model is Claude Fable 5.

What Fable 5 actually is

Claude Fable 5 is Anthropic’s first generally available Mythos-class model. It shares the same underlying model family as the restricted Claude Mythos 5, but adds stronger safeguards for high-risk domains (cybersecurity, biology, chemistry, model distillation).

Fable 5 benchmarks vs Opus 4.8

The benchmark gap is the clearest evidence that Fable 5 is not just an incremental Opus update.

Three patterns stand out:

1. The agentic coding gap is massive. Fable 5’s +11-point lead on SWE-bench Pro over Opus 4.8 is the kind of difference that changes whether an agent finishes a multi-file migration or stalls halfway through.
2. FrontierCode Diamond more than doubles. On production-codebase standards for maintainable agentic coding, Fable 5 scores 29.3% vs Opus 4.8’s 13.4%. Anthropic says Fable 5 leads even at medium reasoning effort.
3. Sonnet 5 closes the gap from below. Sonnet 5’s 63.2% on SWE-bench Pro is close to Opus 4.8’s 69.2%, which means the value proposition of Opus 4.8 is increasingly about accuracy on the hardest tasks rather than a broad capability moat.

Treat vendor-published benchmarks as directional. Validate on your own workload before committing budget. For the full benchmark-by-benchmark breakdown, see Claude Fable 5 vs Opus 4.8.

Fable 5 pricing in detail

The cost strategy is simple: do not use Fable 5 for everything. Use it when the task is hard enough that fewer iterations, better planning, or a higher first-pass success rate pays for the higher token price.

The Fable 5 data-retention caveat

This is the biggest enterprise consideration. Claude Fable 5 is a covered model with a 30-day data-retention requirement. It is not available under zero data retention.

If your organization requires ZDR, Opus 4.8 or Sonnet 5 may remain the better choice until policies catch up. For individual users and teams without strict retention constraints, this is usually acceptable.

GitHub’s Copilot changelog also notes that Fable 5 prompts and outputs may be retained for up to 30 days for safety classifier operation, and that retained data is not used to train Anthropic’s models.

Claude Opus 4.8: the actual latest Opus

Claude Opus 4.8 is the newest and most capable model in the Opus line. It was released on May 28, 2026, and remains the recommended default for complex agentic coding and enterprise work when Fable 5’s price or data-retention requirements are not justified.

What Opus 4.8 improved over Opus 4.7

Opus 4.8 was not a minor patch. Anthropic’s announcement highlighted:

• Stronger coding and agentic-task performance across long-running work
• Better consistency and autonomy on multi-step professional tasks
• Effort control — the effort parameter lets you trade depth for speed and cost
• Claude Code dynamic workflows — better planning and execution in coding agents
• Cheaper fast mode — a lower-latency option for time-sensitive work

Customer feedback from the launch page reinforces the pattern: Opus 4.8’s biggest gains are in judgment and self-correction. It asks better questions, catches its own mistakes, and pushes back on unsound plans more reliably than Opus 4.7.

When Opus 4.8 beats Fable 5

Opus 4.8 is not obsolete. It is the better choice when:

For the full side-by-side, read Claude Fable 5 vs Opus 4.8.

Claude Sonnet 5: the budget disruptor

Claude Sonnet 5, released June 30, 2026, is the model that makes the “should I use Opus?” question harder. It brings near-Opus performance at a Sonnet price.

Sonnet 5’s benchmark story is strong: 85.2% on SWE-bench Verified, 63.2% on SWE-bench Pro, 80.4% on Terminal-Bench 2.1, and 84.7% on BrowseComp. For most everyday coding, research, and professional workflows, Sonnet 5 is now the default recommendation.

The practical model-routing strategy:

1. Sonnet 5 for default agent and coding work.
2. Opus 4.8 when accuracy matters more than cost.
3. Fable 5 for the hardest autonomous work where failure is more expensive than token usage.

For the full breakdown, read the Claude Sonnet 5 guide.

Claude Opus 5 vs the competition: what people actually compare

Many “Claude Opus 5” searches are really cross-provider comparisons. Here is how the Claude lineup maps to the competition.

For the detailed head-to-head, see GPT-5.6 Sol vs Claude Fable 5. The short version: GPT-5.6 Sol is cheaper on short-context API calls ($5/$30 vs $10/$50), but Fable 5 leads on SWE-bench Pro and includes the full 1M context at its standard rate, while GPT-5.6 applies a 2x input surcharge above 272K tokens.

Will there be a Claude Opus 5?

Anthropic has not confirmed or denied a future Opus 5. Here is what we can infer from the current trajectory:

Arguments for an Opus 5 eventually:

• The Opus brand is well-established and widely recognized
• Opus 4.8 remains the recommended default for complex coding and enterprise work
• Not every user needs Mythos-class capability or can accept Fable 5’s data-retention requirements
• A future Opus 5 could bring Fable-class improvements to the Opus tier at a lower price

Arguments against a traditional Opus 5:

• Anthropic created the Mythos tier specifically to separate the highest capability from the Opus line
• The naming structure (Fable, Mythos) suggests a long-term brand strategy beyond version numbers
• Fable 5 already occupies the “next step above Opus” position
• Anthropic may continue iterating on Opus 4.x (4.9, 4.10) rather than jumping to 5

Most likely scenario: Anthropic continues the Opus 4.x line for the premium-but-not-frontier tier, while Fable/Mythos handles the absolute top. If an “Opus 5” ever ships, it would likely be a rebranding of a future Mythos-class model for broader availability — similar to how Fable 5 brought Mythos capability to general users.

This is speculation, not a roadmap. Anthropic has not published a model roadmap.

Which Claude model should you use?

The best metric is cost per successful task, not price per token. A model that finishes a task in one pass at $10/MTok can be cheaper than a model that needs three retries at $3/MTok.

How to access each Claude model

Plan packaging and capacity can change. Fable 5’s included subscription allowance for Pro, Max, Team, and selected Enterprise plans ran through July 7, 2026; continued use after that date may require usage credits. Check your account UI for current entitlements.

FAQ