Scan: Next OpenAI GPT Luna (5.7+) released by September 30, 2026?

1. Planner Fable 5.1 · Fermi decomposition + tool plan

Tools chosen: polymarket_direct polymarket_related kalshi_related article_search claude_news gdelt_news wikipedia

Sub-questions (Fermi decomposition)

What is the current Polymarket price and recent trend for this market, and what is the volume?
When was GPT-5.6 Luna released, and what has OpenAI's cadence been between successive minor GPT versions (5.5 → 5.6, etc.) over the past year?
Is 'Luna' a recurring model family/branding tier at OpenAI (alongside Sol, Terra, Astra) that would carry forward to version 5.7+, or a one-off name that might be dropped?
Are there any credible leaks, roadmap statements, or official OpenAI communications about GPT-5.7, GPT-6, or the next Luna-branded model and its expected release timing?
How many weeks/months remain until Sept 30, 2026 relative to the typical gap between OpenAI model releases?
What do related Kalshi and Polymarket markets (e.g., GPT-6 release, next OpenAI model by date) imply about the probability of a new version by Q3 2026?

Planner reasoning

This is a Polymarket question about whether OpenAI ships a GPT model named 'Luna' at version 5.7+ by Sept 30, 2026. Key drivers are OpenAI's release cadence (how often minor versions ship), whether 'Luna' is a persistent product line or a one-off codename, and any leaks/announcements about GPT-5.7 or GPT-6. The market price is the primary anchor, and cross-venue Kalshi markets on OpenAI releases give an independent read.

2. Research Parallel tool calls

tool	status	time	items	summary
polymarket_direct	OK	0.4s	1	## This Market's Polymarket Data Next OpenAI GPT Luna (5.7+) released by September 30, 2026? - Current price (probability): 90.00% - 7-day price change: +0.50% - 30-day price change: +54.00% - Total volume: $17,432 (USD notional) - Price range: 36.00% - 91.50% - Data points: 30 days
polymarket_related	OK	3.0s	0	Scanned 100 active Polymarket markets, kept 0 matches. keyword 'OpenAI': 0 markets \| keyword 'GPT-5.7': 0 markets \| keyword 'GPT-6': 0 markets \| keyword 'Luna': 0 markets
kalshi_related	OK	2.8s	1	1 related markets / summaries. keyword 'OpenAI': ok \| keyword 'GPT-6': no matches \| keyword 'GPT-5.7': no matches
article_search	OK	0.7s	19	Fetched 19 unique articles across 4 queries (lookback=90d). 'OpenAI GPT-5.6 Luna release': 10 articles \| 'OpenAI GPT-5.7 release date': 10 articles \| 'OpenAI GPT-6 launch timeline': 10 articles \| 'OpenAI Luna Sol Terra Astra model naming': 10 articles
claude_news	OK	23.6s	12	Based on my research, here are the key findings: - Luna is a confirmed recurring tier name: OpenAI's naming convention now uses celestial/geographic tier names alongside version numbers — GPT-5.6 launched July 9, 2026 as a family of three tiers: GPT‑5.6 spans three model tiers: Sol, our flagsh
gdelt_news	OK	156.1s	20	GDELT: 20 articles across 3 queries (lookback=60d). 'OpenAI GPT-5.7': 10 hits \| 'OpenAI GPT-6': error GDELT rate-limited after retries (429) \| 'GPT Luna OpenAI': 10 hits
wikipedia	OK	0.2s	2	Fetched 2 Wikipedia entries (0 missing pages).

3. Evidence Brief Sonnet 5 · 6027 chars

# Event Will OpenAI publicly release a "GPT Luna" model versioned 5.7 or higher (or an equivalent Luna-branded tier under a new whole-number generation, e.g. GPT-6 Luna) by September 30, 2026? # Outcomes to forecast - Yes - No # Kalshi market anchor No Kalshi-direct price returned for this specific ticker in the raw data; the paired Polymarket contract (same event) trades at **90.00% YES**, up from ~36% a month ago (+54% 30-day move), with only modest 7-day movement (+0.5%). Volume is thin ($17.4K total), so the price may be sensitive to a small number of traders. Treat 90% as the closest available consensus anchor; no OpenAI-specific Kalshi related markets were found (only tangential OpenAI IPO/regulatory markets). # Sub-question answers 1. **Polymarket price/trend/volume** — 90% YES, 7d +0.5%, 30d +54%, total volume $17,432, range 36%-91.5% over 30 days [polymarket_direct]. 2. **GPT-5.6 Luna release date & cadence** — GPT-5.6 (Sol/Terra/Luna family) announced late June 2026, briefly restricted for government review, then publicly released July 9, 2026 [CNBC, claude_news]. Prior cadence: GPT-5 (Aug 7, 2025) → GPT-5.5 (Apr 23, 2026) → GPT-5.6 (Jul 9, 2026) → GPT-6 Astra (Sep 3-4, 2026) — roughly 2-4 month gaps, with the 5.6→6 gap being unusually short (~2 months) [claude_news, wikipedia, gdelt]. 3. **Is Luna a recurring branding tier?** — Yes, confirmed: Luna is OpenAI's "fastest/most affordable" tier name, paired with Sol (flagship) and Terra (mid-tier) within the GPT-5.6 generation [claude_news, OpenAI blog]. However, GPT-6 Astra launched as a single flagship model (aka "GPT-6 Pro" in ChatGPT) with no confirmed Sol/Terra/Luna sub-tier structure reported yet [claude_news]. 4. **Leaks/roadmap for GPT-5.7 or next Luna** — No GPT-5.7 was ever released; OpenAI skipped it entirely and jumped to GPT-6 Astra [claude_news, orcarouter.ai]. No confirmed announcement of a "GPT-6 Luna" or Astra-Luna variant as of research cutoff (~Sept 8, 2026 articles). 5. **Time remaining vs. cadence** — From GPT-6 Astra's Sept 3-4, 2026 launch to the Sept 30, 2026 deadline is under 4 weeks — much shorter than the historical 2-4 month gap between major/point releases, meaning a new Luna-tier launch this fast would be atypical unless bundled shortly after Astra (as Sol/Terra/Luna were bundled together on July 9). 6. **Related Kalshi/Polymarket markets** — No direct GPT-6/GPT-5.7 Kalshi or Polymarket markets found (0 matches on keyword scans) [polymarket_related, kalshi_related]. Only tangential OpenAI markets exist (IPO race vs. Anthropic at 94%, US equity stake market at 19%), which don't bear directly on release timing. # Key facts (high-confidence, factual) 1. [CNBC] GPT-5.6 Sol/Terra/Luna family launched publicly July 9, 2026, after a ~2-week government-requested restricted rollout. 2. [claude_news/Wikipedia] GPT-6 Astra released Sept 3 (approved users) / Sept 4, 2026 (general availability) — before question close. 3. [claude_news] GPT-5.7 was never released as a standalone version; OpenAI skipped it in favor of GPT-6 Astra. 4. [gdelt/CNBC] GPT-5.6 Luna pricing was cut 80% on July 30, 2026, and became the default free-tier model in early August 2026 — indicating heavy post-launch investment in the Luna tier under the 5.6 generation, not evidence of a new ≥5.7 Luna model. 5. [claude_news] No confirmed reporting that GPT-6 Astra has spawned Sol/Terra/Luna-style sub-tiers as of Sept 8, 2026 (latest article found). # Cross-market signals - Kalshi related: No OpenAI-model-specific Kalshi markets found; unrelated IPO/regulatory markets don't inform release timing. - Polymarket: Same-event price at 90% YES, with a sharp run-up (+54% in 30 days) likely coinciding with/following GPT-6 Astra's Sept 3-4 launch — suggesting traders may be interpreting Astra's release (or expected imminent Luna-tier companion) as satisfying/likely to satisfy resolution criteria. - Sportsbook implied: N/A. # Analyst opinions and speculation - claude_news synthesis is skeptical: concludes that as of research cutoff, no qualifying "Luna ≥5.7" model has shipped, and standalone Astra (without Luna branding) likely does not qualify under a strict reading — calling a further Luna-branded release by Sept 30 "unlikely unless OpenAI quickly ships budget/Luna variants of Astra." - Historical pattern (5.6 family: Sol/Terra/Luna launched together) suggests if OpenAI extends tiering to GPT-6, a Luna variant could plausibly follow within weeks of Astra's Sept 3-4 launch — this may explain the Polymarket price surge, but no direct confirmation exists in retrieved sources. # Directional lean per outcome - **Yes**: Polymarket at 90% and rising; OpenAI has a demonstrated pattern of bundling Luna-tier releases with flagship launches (5.6 family) and moved fast from Astra's announcement to public rollout; ~4 weeks remain, and OpenAI's recent cadence has accelerated (5.6→6 in ~2 months). - **No**: No sourced confirmation of any Luna-branded GPT-6/Astra variant existing or imminent; GPT-6 Astra shipped as a solitary flagship without tiers per available reporting; strict resolution criteria require explicit "Luna" naming + version ≥5.7, which historically OpenAI skipped (5.7) entirely; short remaining window (~3-4 weeks) is tight for a brand-new tier launch. # Gaps / unknowns - No direct confirmation whether OpenAI has announced or plans a Luna-tier variant of GPT-6 Astra before Sept 30, 2026 — the single most decisive missing fact. - No Kalshi-direct YES price was returned for this ticker (only Polymarket data available as anchor). - Research cutoff appears to be ~Sept 8, 2026; three weeks of potential news (up to close) not captured. # Calibration anchors - Polymarket YES price for this exact event: 90% (anchor, thin volume ~$17K). - Precedent: OpenAI bundled Sol/Terra/Luna tiers simultaneously in July 2026, suggesting tier launches can occur in tight windows once a flagship (e.g., Astra) ships — supports moderate-to-high probability but not certainty absent confirmed reporting.

4. Forecasters Fable 5.1 + GPT-5.6 Sol · in parallel

claude-fable-5-1

52% conf

Yes	85%
No	15%

The paired Polymarket contract sits at 90% YES after a +54% surge that coincided with GPT-6 Astra's Sept 3-4 release, implying traders largely believe a 5.7+ generation release (Astra, or an imminent Luna-tier companion) satisfies the market. OpenAI's July 2026 precedent of bundling Sol/Terra/Luna tiers and its accelerated cadence (5.6→6 in ~2 months) make a GPT-6 Luna variant within the remaining ~4 weeks plausible. However, as of the ~Sept 8 research cutoff no Luna-branded GPT-6 variant is confirmed, and a strict reading requiring explicit 'Luna' naming leaves genuine resolution risk, so I shade slightly below the thin-volume 90% anchor.

gpt-5.6-sol

56% conf

Yes	68%
No	32%

The closest consensus anchor is Polymarket at 90% YES, but its thin $17.4K volume and abrupt 54-point monthly rise make it vulnerable to traders treating the GPT-6 Astra launch itself as sufficient. Strictly, Astra does not qualify without explicit Luna branding, GPT-5.7 was skipped, and no GPT-6 Luna announcement was confirmed as of September 8. The prior Sol/Terra/Luna bundle and OpenAI’s accelerating release cadence make a companion Luna tier within the remaining three weeks plausible, but the sub-four-week window is substantially shorter than the usual 2–4 month cadence. I therefore remain above 50% but discount the market materially for the naming requirement and absence of direct launch evidence.

5. Devil's Advocate Sonnet 5 · red-team critique

Both forecasters anchor heavily on Polymarket's 90% despite explicit brief warnings that thin volume ($17.4K) makes it "sensitive to a small number of traders" and that the price surge likely reflects traders conflating Astra's launch with satisfying resolution criteria—yet neither forecast interrogates whether that conflation is actually justified under strict resolution wording, just gesturing at "resolution risk" without quantifying it.

Key fact #3 states unambiguously "GPT-5.7 was never released as a standalone version; OpenAI skipped it in favor of GPT-6 Astra" — this is near-decisive evidence that the specific 5.7 threshold is dead, yet both forecasts treat this as a minor caveat rather than recognizing the question now hinges entirely on the disjunctive "Luna-branded tier under a new whole-number generation," a much narrower and unconfirmed condition (fact #5: "No confirmed reporting that GPT-6 Astra has spawned Sol/Terra/Luna-style sub-tiers").

Neither forecaster adequately weighs sub-question 5 (time remaining vs. cadence): the historical Sol/Terra/Luna bundle happened simultaneously with the flagship (not weeks later), meaning if Astra didn't ship with a Luna tier on Sept 3-4, the base rate for a standalone Luna tier appearing in the following ~4 weeks is actually low, not "plausible" — this is a reference-class error where they treat "cadence has been fast" as supporting Yes when the specific bundling precedent argues the window for a delayed Luna companion may have already closed.

Both forecasts fail to consider the tail scenario where OpenAI's tiering strategy shifted with GPT-6 (single flagship model "Astra"/"GPT-6 Pro" with no sub-tier structure at all per available reporting) — i.e., that Luna branding could be permanently retired rather than merely delayed, which would argue for a probability closer to 30-40% than 68-85%.

The two forecasts diverge substantially (0.85 vs 0.68, a 17-point gap) on essentially the same evidence set, suggesting at least one is miscalibrated; Forecast 1's rationale doesn't sufficiently justify staying so close to the raw 90% anchor given it explicitly acknowledges "genuine resolution risk," while Forecast 2's discount to 0.68 seems more consistent with the brief's skeptical analyst synthesis but neither anchors to a specific base rate for "unconfirmed feature ships within 3-4 weeks of speculation."

Neither forecast mentions the March-to-3-week research gap (cutoff ~Sept 8, close Sept 30) as a source of irreducible uncertainty that should widen error bars rather than just cause a modest haircut — three weeks of unobserved news could easily resolve this decisively in either direction, arguing for a probability nearer 50-60% reflecting the coin-flip nature of unresolved binary information, not the 68-85% range both landed in.

6. Aggregator Fable 5.1 · final calibrated forecast

76% (54% confidence)

Yes: 76% No: 24%

The only market anchor is Polymarket at 90% YES, but it is thin ($17K) and its surge coincided with GPT-6 Astra's launch, which under a strict reading does not itself qualify (no Luna branding, and GPT-5.7 was skipped entirely). The critique correctly notes that the Sol/Terra/Luna precedent was a simultaneous bundle, so Astra shipping solo weakens the case for a Luna companion within ~4 weeks, and a tiering-strategy shift is a real tail. Still, the price has held at ~90% for a week after Astra's release rather than collapsing, suggesting informed traders either expect an imminent budget/Luna tier or believe the resolution wording is satisfied; three weeks of unobserved news also leave meaningful room for a qualifying launch. I discount the anchor materially for the naming/strict-resolution risk and absence of confirmed reporting, landing between the two forecasts but closer to Forecast 1 given market persistence.

Pipeline Timing

Total pipeline time: 268.5s

Per-tool research timings shown in the Research section above.