What Is OpenAI Spud?
OpenAI's next frontier model — internally codenamed "Spud" — completed pretraining on approximately March 24–25, 2026, according to reporting by The Information. CEO Sam Altman told employees in an internal memo that the company expects a "very strong model" within "a few weeks" that can "really accelerate the economy." No public benchmark results, parameter count, or final product name have been disclosed. Whether it ships as GPT-5.5 or GPT-6 is currently unknown.
What is confirmed: training is done, organizational changes are already in motion, and one of OpenAI's most-used products was shut down to free the compute to get there.
What OpenAI Did to Ship Spud Faster
The Spud pretraining announcement came bundled with a cluster of decisions that reveal how seriously OpenAI is treating this release cycle.
Sora is dead. OpenAI shut down its Sora video generation app, its mobile app, and its API entirely. Altman told staff that Sora was "a drag" on resources. Every GPU previously running Sora video inference has been redirected toward Spud training and evaluation. A planned partnership with Disney, reported earlier in 2026, is now dead with it. The Sora team has been reassigned to world simulation research targeting robotics applications.
The product org was renamed. Fidji Simo's division at OpenAI — previously called "Product" — was officially renamed "AGI Deployment." Simo told employees the company had been "spreading efforts across too many apps" and could not afford "side quests" while Anthropic was winning enterprise customers. The rename is not just symbolic: it signals that OpenAI's internal framing around Spud is AGI-adjacent, not standard product.
Altman restructured leadership. AI safety reporting now runs through CRO Mark Chen's research division. Security reports to president Greg Brockman, who leads what's internally called the "Scale" division. Altman himself is focused on fundraising, supply chain, and data center construction — not day-to-day product decisions.
Spud may also serve as the foundation for OpenAI's planned desktop "superapp," which would combine ChatGPT, the coding agent Codex, and the browser Atlas into a single interface.
The IPO Pressure Behind Spud
Spud is not shipping in a vacuum. OpenAI is preparing for what could be the largest technology IPO in history, and the model launch matters for the company's valuation narrative.
OpenAI crossed $25 billion in annualized revenue at the end of February 2026, up from $21.4 billion at year-end 2025 and roughly $6 billion at the end of 2024 — a 4x increase in 14 months. Despite that trajectory, Anthropic has been growing faster in enterprise market share, and the gap between the two companies has compressed significantly.
Among U.S. businesses tracked by Ramp Economics Lab, Anthropic's share of combined OpenAI-plus-Anthropic enterprise spend went from roughly 10% at the start of 2025 to over 65% by February 2026. That is the competitive context inside which Spud is being deployed.
OpenAI CFO Sarah Friar has told associates the company is targeting a 2027 listing, while some advisers believe a late 2026 regulatory filing is possible. A frontier model launch in Q2 2026 — timed to coincide with investor attention ahead of an S-1 — is not an accident.
| Metric | Figure | Source / Date |
|---|---|---|
| OpenAI annualized revenue | $25B | The Information, Feb 2026 |
| OpenAI valuation (last round) | $840B post-money | February 2026 funding round |
| Anthropic enterprise spend share | 65% of OpenAI+Anthropic combined | Ramp Economics Lab, Feb 2026 |
| Spud pretraining completed | ~March 24–25, 2026 | The Information, March 25 |
| Expected public release | "Within weeks" of March 25 | Sam Altman internal memo |
What Spud Actually Is (and What We Don't Know)
Pretraining is the first — and most computationally expensive — phase of model development. It is not the finish line. After pretraining, models go through post-training alignment, safety evaluations, red-teaming, and infrastructure scaling before any public release. Altman's memo signals the hardest compute work is done; it does not mean Spud ships tomorrow.
Spud is being trained at the Stargate facility in Abilene, Texas — the first Stargate site — using over 100,000 H100 GPUs. The parameter count has not been disclosed, nor has whether it is a reasoning model, non-reasoning, or something with a new architecture entirely.
OpenAI employees have hinted that Spud contains a capability that is "very different from what we've seen before," though no specifics have been shared publicly.
What has been confirmed or credibly reported, versus what remains unknown:
| Status | Detail |
|---|---|
| ✅ Confirmed | Pretraining complete as of ~March 24, 2026 |
| ✅ Confirmed | Release expected "within weeks" per Altman |
| ✅ Confirmed | Sora compute redirected to Spud |
| ✅ Confirmed | Org renamed to "AGI Deployment" |
| ❌ Unknown | Final product name (GPT-5.5 vs GPT-6) |
| ❌ Unknown | Parameter count or architecture details |
| ❌ Unknown | Pricing tiers |
| ❌ Unknown | Benchmark scores |
What This Means for Developers
Don't Rebuild Your Stack Yet
Pretraining completion is a milestone, not a deployment. Completing pretraining is roughly equivalent to finishing a building's structure — important, but there's still significant work before anyone moves in. Post-training alignment, capability evaluations, and infrastructure rollout typically take weeks to months for a frontier-class model. Developers relying on the OpenAI API for production workloads should continue on current models (GPT-5.4 and its variants) until Spud's capabilities and pricing are confirmed.
The Superapp Matters More Than the Model Name
Whether Spud ships as GPT-5.5 or GPT-6 is the wrong question. The strategic move is the superapp: merging ChatGPT, Codex, and the Atlas browser agent into a single desktop interface. OpenAI is currently prioritizing the coding and enterprise products space, particularly to close the gap with competitor Anthropic, which has been gaining traction with Claude Code. If the superapp ships alongside Spud, it changes the developer workflow calculus — less about which model, more about which integrated platform.
The Anthropic Race Is Now Public
Both Anthropic (Claude Mythos, training complete, invite-only cybersecurity rollout) and OpenAI (Spud, pretraining complete, expected release "within weeks") are confirming frontier-level models in the same week. This is not coincidence. Both companies are moving toward high-stakes milestones — OpenAI toward an IPO, Anthropic toward expanded enterprise deals — and model launches are central to both narratives.
For developers choosing between platforms right now: both ecosystems are on the verge of significant capability updates. If your evaluation criteria include cybersecurity or autonomous coding agents, the Claude Mythos early access program is the more concrete near-term opportunity. If your criteria include platform integration and a broad consumer user base, watch Spud's launch closely for pricing and API access details.
FAQ
What is OpenAI Spud?
Spud is the internal codename for OpenAI's next frontier model. Pretraining completed around March 24–25, 2026. Sam Altman told employees it is a "very strong model" that can "really accelerate the economy," with release expected within weeks. Its final product name has not been confirmed.
Is Spud GPT-6?
Unknown. OpenAI has not confirmed whether Spud will ship as GPT-5.5, GPT-6, or under a different branding entirely. Given that GPT-5.3 and 5.4 were released as intermediate updates, GPT-5.5 is the more conservative estimate, but OpenAI has not ruled out a version-number jump.
Why did OpenAI shut down Sora?
Altman told staff that Sora was "a drag" on compute resources. The GPU capacity previously used for video generation has been redirected to Spud training and evaluation. A planned Disney content deal is also now dead. The Sora team is moving to world simulation research for robotics.
When will Spud be released publicly?
Sam Altman's internal memo said "a few weeks" from approximately March 25, 2026, placing a potential release window in April–May 2026. No official launch date or developer preview access program has been announced.
How does Spud relate to OpenAI's IPO?
OpenAI crossed $25 billion in annualized revenue in February 2026 and is preparing for a public listing potentially targeting a $1 trillion valuation. A strong frontier model launch directly supports that narrative. The renaming of the product org to "AGI Deployment" signals that Spud is being positioned as a foundational IPO-era asset, not just another release.
Next step: Watch The Information and OpenAI's developer blog for the first Spud benchmark disclosures — that data will be more useful than any pre-launch memo for evaluating whether it actually closes the gap with Claude Opus 4.6 on coding and reasoning tasks.
Sources: The Information, March 25, 2026 (https://www.theinformation.com) · The Decoder, March 24, 2026 (https://the-decoder.com/openai-ceo-sam-altman-reportedly-teases-a-very-strong-model-internally-that-can-really-accelerate-the-economy/)