Unverified 'GPT Terra' Model Surfaces on OpenRouter With 1.05M-Token Context, No OpenAI Confirmation

What's being reported

A listing on OpenRouter's model catalog shows an entry called "OpenAI GPT Terra Latest," accessible via the API identifier ~openai/gpt-terra-latest. According to the listing, this is a rolling alias that always redirects to whatever is the newest model in a "GPT Terra" family — a name that does not correspond to any model line OpenAI has publicly announced.

The tilde (~) prefix on OpenRouter's slug is notable: OpenRouter uses this convention for community-tracked or provisional listings rather than entries that a model provider has directly submitted and confirmed. There is no press release, blog post, or API documentation from OpenAI referencing "GPT Terra" as of this writing.

The specs, as listed

Per the OpenRouter page, the model carries the following specifications:

• Context window: 1,050,000 tokens (roughly 1.05 million)
• Pricing: $2.00 per million input tokens, $12.00 per million output tokens
• Modality: text, image, and file input → text output
• Access: OpenRouter API only, under the alias ~openai/gpt-terra-latest

No benchmark scores, parameter count, or training data cutoff date are listed. No system card or safety evaluation has been published alongside the entry.

Why this needs scrutiny

OpenAI's confirmed model families are named GPT-4, GPT-4o, GPT-4.1, GPT-5, and the o-series reasoning models (o1, o3, o4-mini). "GPT Terra" does not match any of these naming conventions, and OpenAI has not listed it on its own API pricing or documentation pages. It is possible this listing reflects an internal codename that leaked into a routing layer, a placeholder used for testing infrastructure, or a mislabeled entry on OpenRouter's end. It is also possible OpenRouter has early access to an unannounced model under an aggregator agreement, which has happened before with other providers.

Without a statement from OpenAI, none of these explanations can be confirmed. The 1.05 million token context window would be larger than any context window OpenAI currently offers publicly (GPT-4.1 tops out at 1 million tokens in its long-context variant), which makes independent verification especially important before treating the specs as final.

What this means

Treat this listing as unconfirmed until OpenAI issues its own documentation or announcement. If accurate, a 1.05M-token context window at $2/$12 per million tokens would undercut GPT-4.1's long-context pricing while extending context length further, which would matter for document-heavy and long-conversation use cases. But aggregator listings occasionally surface placeholder or test entries before a provider is ready to talk publicly, and "latest" aliases in particular are prone to being repointed or removed without notice. Builders evaluating this model via OpenRouter should expect the specs, pricing, or even the model's availability to change without warning until OpenAI confirms the release directly.