OpenAI has officially confirmed it is developing a new flagship AI model internally codenamed ‘Spud’. Following the company’s historic $122 billion fundraise earlier this week, President Greg Brockman teased the model during a podcast appearance, describing it as the culmination of “two years worth of research” and a significant milestone on the path to Artificial General Intelligence (AGI).
While OpenAI has recently focused on incremental updates (releasing GPT-5.4 just last month), ‘Spud’ is being positioned as a “generational leap” in capability rather than a minor version increment.
1. The “Big Model Smell”
In a nod to the “new car smell” phrase, Brockman used the term “big model smell” to describe the tactile jump in intelligence users should expect.
- Intent Inference: The core focus of Spud is reducing the “friction of explanation.” Brockman noted that the model is designed to better infer user intent and context, significantly minimizing the need for repetitive prompting or detailed “hand-holding” to get a correct result.
- Intuitive Interaction: The goal is for users to interact with the AI as they would a highly competent human colleague—without having to “overthink” how they phrase their requests.
2. Development Status & Release Rumors
Reports indicate that ‘Spud’ has recently completed its primary training phase at the Stargate supercomputing facility.
| Phase | Estimated Status (April 2026) |
| Pre-training | ✅ Completed (utilizing over 100,000 H100 GPUs) |
| Post-training/RLHF | 🟡 In Progress |
| Red-Teaming/Safety | 🟡 Starting |
| Public Release | 📅 Targeted for Late 2026 |
While the official product name remains unconfirmed, industry analysts are debating whether it will launch as GPT-6 or a new “top-of-stack” tier in the o-series reasoning models.
3. Strategic Realignment: The “Sora” Sacrifice
To free up the massive compute resources required for Spud’s final fine-tuning and safety evaluations, OpenAI has made the controversial decision to shut down its Sora video generation platform.
- Focusing on Logic: OpenAI is reportedly scaling back experimental “creative” tools to prioritize foundational reasoning and agentic reliability.
- Enterprise Priority: The shift is aimed at winning back the enterprise market, where rivals like Anthropic (with their leaked “Mythos” model) have recently challenged OpenAI’s dominance in coding and cybersecurity.
4. Capabilities: Reasoning vs. General Purpose
Early signals suggest Spud will blend the deep reasoning of the ‘o’ series with the speed and multimodal fluidity of the ‘GPT’ series.
- High-Horizon Reasoning: Expected to excel at tasks requiring hundreds of steps, such as managing a full software development lifecycle or complex scientific research.
- Economy Accelerator: Sam Altman has reportedly told employees that Spud is a “very strong model” that could “really accelerate the global economy” by automating high-value cognitive workflows.
5. Potential Pricing Tiers
Given its “frontier” status, Spud is expected to sit at the absolute top of OpenAI’s pricing hierarchy.
- API Access: Likely to launch at a premium price point, potentially higher than the current o3 or GPT-5.4 rates.
- ChatGPT Integration: It remains unclear if Spud will be available on the standard “Plus” ($20/mo) tier or if it will require a new “ChatGPT Pro” or “Ultra” subscription.
