GPT-6 release date

The market’s hierarchy points to a specific timing thesis: OpenAI has a successor moving through internal evaluation, yet a July or early-August public release would require an unusually compressed path from testing to broad availability. The clearest evidence is the curve itself. The probability reaches only 6.4% by August 14, rises to 42% by August 31, then reaches 76% by September 30 and 87.5% by year-end.

September carries the launch thesis

The 34-point gap between August 31 and September 30 contains the largest block of implied release probability among the listed monthly deadlines. September also gained five percentage points over the latest 24-hour period, while December rose only 0.5 points. Market inference: recent positioning shifted expected timing toward the third quarter without materially changing confidence in a release sometime during 2026.

That distinction matters because a broad change in beliefs about whether GPT-6 launches this year should affect the December contract more strongly. Concentrated movement in September instead fits a scheduling interpretation, potentially involving an anticipated announcement, staged rollout or completion of evaluations. The supplied record contains no official GPT-6 date, so attributing the move to a particular leak, event or trader would be unsupported.

An unnamed test model supports the curve, with conditions

OpenAI said on July 21 that an “even more capable” pre-release model was under internal testing during a security-evaluation incident, alongside GPT-5.6 Sol. This is the strongest factual basis for expecting another model release: development has progressed far enough for internal evaluation. It does not establish the model’s commercial name, release scope or timetable.

The market’s September concentration therefore rests on hidden assumptions. The tested model must become GPT-6, complete safety and product evaluations quickly, and reach the general public under terms that satisfy resolution. A limited research preview, private partner access or a differently branded release could provide evidence of technical progress while leaving a GPT-6 contract unresolved. Exact naming has particular weight because OpenAI’s current public line uses incremental labels including GPT-5.5 and GPT-5.6.

GPT-5.6 gives OpenAI room to delay another generation

As of July 21, OpenAI’s official newsroom still identified GPT-5.6 as its newest public GPT release. The July 9 product announcement made GPT-5.6 available through ChatGPT, Codex and the API, giving OpenAI a recently launched model across its principal distribution channels. The newsroom contained no GPT-6 announcement in its recent product flow.

A June 25 announcement adds a competing causal explanation for a slower rollout. OpenAI expanded its personal-finance experience to US Plus and Pro users while crediting GPT-5.5 advances for stronger reasoning on complex, context-dependent tasks. This shows that product expansion and monetization can continue through existing model families. Market inference: OpenAI may have operational incentives to improve adoption, capacity and specialized products before introducing a new generation name.

Safety evaluation can signal progress and extend the schedule

The security incident disclosed during model evaluation cuts in both directions. Internal access to a more capable pre-release system confirms work beyond GPT-5.6. The involvement of a security evaluation also shows that release readiness includes controls outside benchmark performance. Additional investigations, access restrictions or repeated testing could weaken the late-August and September thesis even if the underlying model remains technically advanced.

Evidence supporting September would include an OpenAI announcement naming GPT-6, documentation entering the API or ChatGPT release channels, or a dated public rollout plan. Evidence favoring later deadlines would include another GPT-5.x launch, continued references to the tested system as pre-release, expanded safety work, or availability limited to evaluators. A differently named flagship would directly challenge the assumption that technical succession guarantees GPT-6 branding.

Rule wording is the main non-model risk

The supplied market record contains a material inconsistency: the event lists 2026 deadline outcomes, while the displayed resolution criterion refers to public availability by December 31, 2025, and the listed close date is also December 31, 2025 despite the market being marked open. The multi-timeframe rule says each listed date corresponds to an underlying binary market, which suggests date-specific contracts, but the underlying wording requires clarification.

This matters independently of OpenAI’s release cadence. With $635,250 in volume, $151,550 in open interest and $95,720 in liquidity, the event has meaningful participation, yet those figures cannot resolve ambiguous settlement language. Confirmation of the underlying 2026 criteria would preserve the technology-driven interpretation. A determination that the 2025 date governs would force a reassessment based on contract administration instead of GPT-6 development evidence.

Sources