Uncovering Claude Opus 4.6: User Discontent Rises Over 4.7's Value and Anthropic's Transparency

The recent rollout of Claude Opus 4.7 has sparked significant discussion among users, particularly concerning the perceived disappearance of its predecessor, Opus 4.6, and the overall value proposition of Anthropic's latest offerings. While initial reactions suggested a silent removal of Opus 4.6, user-discovered methods quickly emerged, shedding light on its continued, albeit less prominent, availability.

Accessing Previous Claude Models

Contrary to initial belief, Opus 4.6 has not been entirely removed. Users can still access it by typing specific commands within the chat interface, such as /model claude-opus-4-6. Additionally, for those using the graphical interface, Opus 4.6 (and Sonnet 4.5) can typically be found under a "More Models" section. This discovery provides a workaround for users who preferred the previous versions or found them more suitable for their tasks and budget, despite wishes for it to be more clearly presented in model selection controls.

Value and Cost Concerns with Opus 4.7

Despite its introduction, Opus 4.7 has been met with mixed reception regarding its value for money. Many users report that while it's the newer model, it has effectively halved their usage capacity (e.g., turning a "Max 5x" experience into "Max 2.5x" over a 5-hour period), making previous usage limits significantly more restrictive. This increase in effective cost, without a perceived proportional improvement in performance, has led to frustration and cancellations. The sentiment is that Anthropic's policies are becoming increasingly subscriber-hostile and lack transparency, with some suggesting that higher costs should be accompanied by clearer messaging and potentially optional stricter limits.

The Business Perspective: Profit vs. User Experience

A significant theme emerging from the discussion is the inherent tension between AI model providers' need for profitability and user expectations for consistent service and value. Some argue that actions perceived as user-hostile, like the hidden availability of older models or increased effective pricing, are not malicious but rather a consequence of insufficient compute resources. Companies are building infrastructure as fast as possible, and hard decisions are inevitable when demand outstrips supply.

Conversely, another perspective posits that corporate actions are almost universally driven by profit maximization. In this view, any decision that doesn't overtly prioritize generating more income from fewer computational resources is unlikely to be approved. This perspective suggests that functionality will likely degrade over time for paying users, with providers showcasing peak performance to attract new customers, then gradually reducing it to maximize revenue per unit of work. This phenomenon is often linked to the concept of "enshittification" – where platforms gradually worsen for users to extract more value.

One insightful recommendation suggests that for critical workflows, reliance on public-facing subscription models is inherently unstable. Instead, users should leverage free tiers or initial high-performance offerings to develop infrastructure for models they can control, either on in-house hardware or leased VPS instances, ensuring stable and reliable performance.

Seeking Alternatives

The dissatisfaction with Opus 4.7's cost-performance ratio and Anthropic's perceived lack of transparency has driven many users to explore alternative AI services. While specific recommendations were limited, one user mentioned a "$20 Codex subscription" as a potentially better or equivalent option for usage allowance. The demand for alternatives highlights a growing market where users are actively seeking better value and more transparent pricing structures.