Claude: rollout updates for advanced models and availability

Claude’s 2025 roadmap has delivered a wave of advanced model releases, expanded context windows, and new subscription structures that reshape how individuals and enterprises can access Anthropic’s technology. The most recent rollouts include Opus 4.1, the public beta of Sonnet 4 with a one-million-token context, and revised rate-limit rules for higher-tier customers. Here we detail the latest availability updates and their implications for different user segments.

Opus 4 introduces next-generation reasoning.

The Claude Opus 4 model was released on 22 May 2025, replacing Opus 3 as Anthropic’s most capable reasoning system.

• Access tiers: Available to Pro, Team, Max, and Enterprise users, as well as through AWS Bedrock and Google Vertex AI.

• Context window: Maintains a 200,000-token limit, sufficient for book-length documents and large codebases.

• Improvements: Provides more consistent long-form reasoning, better factual grounding, and stronger coding capabilities compared to prior Opus generations.

Opus 4 has become the reference point for complex enterprise workflows, where accuracy and reliability matter as much as raw speed.

Opus 4.1 upgrade boosts coding and latency.

On 5 August 2025, Anthropic introduced Claude Opus 4.1 as an incremental upgrade.

• Performance gains:

  • +7 percentage points on SWE-bench, reaching over 70% accuracy in coding benchmarks.

  • Faster tool-use response times and improved function-calling stability.

• Context handling: Still limited to 200,000 tokens, with enhanced retrieval across lengthy inputs.

• Deployment: Rolled out simultaneously on the web client, API, and AWS Bedrock.

This update ensures enterprises relying on Opus for software development or data engineering tasks gain measurable efficiency improvements.

Sonnet 4 reaches broader users and expands context.

Alongside Opus 4, Anthropic released Claude Sonnet 4 in May 2025 as the mid-tier balance of speed and reasoning.

• Access tiers: Available across Free, Pro, Team, Max, and Enterprise, making it the most widely deployed model in the Claude family.

• Pricing advantage: Around 30% cheaper than Opus 4 while retaining the same 200,000-token context for most tasks.

• Upgrade path: Gradually replaced Sonnet 3.7, which had been the default for the free tier.

Sonnet 4 now powers everyday usage for millions of free and paid subscribers, serving as the default model for fast interactions.

One-million-token context beta arrives for Sonnet 4.

The most significant capacity upgrade came with the Sonnet 4 one-million-token context beta, rolled out on 11 August 2025.

• Access tiers: Available through API header opt-in, Max plan, and Enterprise accounts.

• Capabilities: Supports workloads such as:

  • Multi-hundred-page legal and compliance documents.

  • Entire software repositories.

  • Complex data analysis pipelines without truncation.

• Beta nature: Offered under the header context-1m-2025-08-11, subject to performance monitoring and quota adjustments.

This rollout makes Claude the first mainstream assistant to publicly test a million-token window, setting a new benchmark for handling expansive data.

Haiku 3.5 remains the fast entry point.

While Opus and Sonnet models saw significant upgrades, Claude Haiku 3.5 continues as the free-tier default and the speed-optimized option.

• Context window: Hard-capped at 48,000 tokens.

• Strengths: Prioritizes low latency and lightweight use cases, ideal for quick responses or mobile usage.

• Role: Serves as a practical entry point for new users before upgrading to Sonnet or Opus.

Haiku’s stability ensures that Anthropic maintains a broadly accessible free tier without compromising higher-end capacity for paid users.

New rate-limit policies change how usage is allocated.

Anthropic announced a rate-limit overhaul on 31 July 2025, set to take effect 28 August 2025.

• Weekly usage caps: Users will be allocated limits on a weekly basis instead of daily resets.

• Separated quotas: Opus tokens are now tracked separately from Sonnet and Haiku usage.

• Max plans:

  • Max 5× plan at $90/month.

  • Max 20× plan at $180/month.

  • Both tiers unlock higher throughput and access to early beta features.

• Enterprise accounts: Maintain custom contracts, with options to purchase on-demand overage capacity at API rates.

These changes aim to provide more predictable usage for enterprises while giving power users more flexibility to scale their workloads.

Comparing Claude’s advanced model availability.

Model	Launch / update	Context window	Key audience	Notable impact
Opus 4	May 2025	200k tokens	Enterprise, Pro+	Highest reasoning depth, complex workflows
Opus 4.1	Aug 2025	200k tokens	Dev & data teams	Faster coding, +7 SWE-bench
Sonnet 4	May 2025	200k tokens	Free → Enterprise	Balanced speed vs reasoning
Sonnet 4 (1M beta)	Aug 2025	1M tokens	Max, Enterprise	Handles books, repos, massive docs
Haiku 3.5	Oct 2024	48k tokens	Free tier	Fastest responses, light tasks

Key takeaways.

• Opus 4.1 refines Anthropic’s most powerful model with stronger coding accuracy and reduced latency.

• Sonnet 4 delivers enterprise-grade context to free and paid tiers, making it the most widely adopted Claude model.

• The 1M-token context beta positions Claude ahead in large-scale document and repository handling.

• Haiku 3.5 continues to anchor the free tier with fast performance.

• Rate-limit changes arriving at the end of August will restructure how Pro, Max, and Enterprise customers consume Claude tokens.

Together, these rollouts mark Claude’s most ambitious expansion year to date, giving both individuals and enterprises more capacity, more reliability, and more choice across the model family.

____________

FOLLOW US FOR MORE.

DATA STUDIOS

datastudios.org