Token Price Index — GPT-5.6 holds the prior-generation price rungs
GPT-5.6 (2026-07-16) holds the exact prior-generation per-token price at two of three tiers: Sol (frontier) $5/$30 = GPT-5.5's rung, Terra (strong) $2.50/$15 = GPT-5.4's rung; Luna (mid) $1/$6 is a new point. Capability up, price rungs held.
GPT-5.6 holds the prior-generation price rungs (Token Price Index, 2026-07-16)
On 2026-07-16 OpenAI released GPT-5.6 in three named variants. Recorded into the MenFem Token Price Index against the price rows already held for the prior generation (verified 2026-07-06), the new prices land on the existing ladder rather than below it.
The observation
| GPT-5.6 variant | Tier | Input / Output (per Mtok) | Prior-generation rung at that tier | Match? |
|---|---|---|---|---|
| Sol | frontier | $5.00 / $30.00 | GPT-5.5 — $5.00 / $30.00 | exact hold |
| Terra | strong | $2.50 / $15.00 | GPT-5.4 — $2.50 / $15.00 | exact hold |
| Luna | mid | $1.00 / $6.00 | GPT-5.4-mini — $0.75 / $4.50 | new point (priced above the prior mid) |
Cached-input prices track the same pattern (Sol $0.50, Terra $0.25, Luna $0.10). All three rows observed 2026-07-16; the prior-generation rows (GPT-5.5, GPT-5.4, GPT-5.4-mini) observed 2026-07-06.
What it shows
- Two of three rungs held to the cent. A capability generation (GPT-5.5 -> GPT-5.6 Sol; GPT-5.4 -> GPT-5.6 Terra) shipped at the identical published per-token price. The price point is a fixed ladder rung that a better model steps onto — not a number that falls per release.
- Luna is the honest exception. The new mid variant at $1/$6 is a new price point, above OpenAI's prior mid rung (GPT-5.4-mini, $0.75/$4.50). So "the rung never moves" is the pattern at the frontier and strong tiers this cycle, not a law — the mid tier got a new, higher rung, not a held one.
Limitations
- One provider, one release event — a pattern data point, not a proven rule. Needs repeated observation across providers/generations before it generalizes.
- Tier assignments (frontier/strong/mid) are the Index's provisional capability buckets pending a benchmark-calibration pass (see research/inference/methodology.md), not OpenAI's own labels.
- Nominal published prices only — actual paid cost varies with caching %, batch discounts, and negotiated enterprise rates not captured here.
Source: MenFem Token Price Index, research/inference/token-prices.csv (rows observed 2026-07-16 and
2026-07-06), each verified against OpenAI's own pricing page,
developers.openai.com/api/docs/pricing. Own dataset —
primary by direct provider-page read.