OpenAI Removes Chat Limits for All Users as GPT-5.6 Luna and Sol Models Roll Out Across Tiers

OpenAI has announced a sweeping update to ChatGPT that removes usage limits on text-based conversations for all users, coinciding with the platform crossing the milestone of 1 billion weekly active users. The overhaul introduces two new model variants: GPT-5.6 Luna for free and Go-tier users, and an upgraded GPT-5.6 Sol for Plus and Pro subscribers, alongside new interface tools that give users direct control over reasoning depth. While text chats are now unlimited, OpenAI clarified that separate rate limits will remain in place for file uploads, image generation, voice interactions, and other multimodal features.

For Free and Go users, GPT-5.6 Luna replaces GPT-5.5 as the default model and serves as the engine behind the newly unlimited text experience. This cohort will also receive a new “Think” button, allowing them to manually activate higher-intensity reasoning for complex queries rather than relying solely on automatic detection. These changes are scheduled to roll out this week, with full unlimited text access and the Think button arriving next week. The move effectively eliminates one of the most visible friction points for non-paying users, signaling OpenAI’s strategy to deepen engagement at massive scale by removing barriers to core functionality.

Plus and Pro users receive a distinct upgrade path. They gain access to an enhanced version of GPT-5.6 Sol (separate from the Codex/Work variant, which remains unchanged), optimized for quick-turnaround tasks including web research, planning, writing, advice, and decision-making. OpenAI describes this model as producing more compact yet robust responses, balancing speed with thoroughness. Crucially, paid users also receive a thinking slider, a granular control that lets them adjust how much computational reasoning the model applies to any given query. This allows power users to dial up depth for multi-step problems or dial it down for straightforward questions, offering a level of customization previously unavailable in consumer-facing AI interfaces.

The performance gains backing these UX changes are substantial. According to OpenAI’s internal evaluations, GPT-5.6 Luna reduces factual errors by 62% compared to GPT-5.5-Instant, while GPT-5.6 Sol achieves a 68% reduction. In an era where hallucination rates remain a primary concern for enterprise and consumer adoption alike, these improvements represent meaningful progress in reliability. Fewer factual errors translate directly to greater user trust, especially as ChatGPT becomes embedded in workflows where accuracy is non-negotiable.

The timing of this update is strategically significant. Crossing 1 billion weekly users places ChatGPT in rarefied air alongside platforms like WhatsApp, Instagram, and TikTok. At this scale, even marginal improvements in retention and session frequency compound into enormous competitive advantages. Removing text limits for free users lowers the barrier to habitual use, potentially converting casual experimenters into daily users who eventually hit multimodal caps and consider upgrading. Meanwhile, the thinking slider and refined Sol model give paying subscribers tangible reasons to maintain their subscriptions beyond mere access they now have better access, with controls that adapt the AI to their specific cognitive needs.

OpenAI’s decision to keep separate limits on files, images, voice, and generation also reveals a deliberate tiering philosophy. Text chat is treated as the foundational utility layer, made universally accessible to sustain network effects and data flywheels. Compute-intensive modalities remain gated, preserving margin structure and incentivizing upgrades without alienating the base. This bifurcation acknowledges that not all tokens cost the same to serve, and that sustainable scaling requires matching resource allocation to user value tiers.

As the updated GPT-5.6 Sol goes live for Plus and Pro users today and Luna rolls out to free users this week, the industry will be watching closely. Unlimited text chat at billion-user scale is an unprecedented infrastructure challenge, and the real-world performance of the Think button and thinking slider will determine whether these features become beloved staples or forgotten novelties. But one thing is already clear: OpenAI is no longer playing defense on usage caps. With GPT-5.6, they are betting that openness, precision, and user agency not artificial scarcity are the true drivers of long-term dominance in the age of ubiquitous AI.