Which current models does Entroly support with verified metadata?

Entroly carries verified public metadata for GPT-5.6 Sol, Terra, and Luna; Gemini 3.6 Flash; Gemini 3.5 Flash-Lite; and discovers installed NVIDIA Nemotron 3.5 Lightning Ollama tags. The request must traverse the corresponding Entroly route.

Does Entroly support Gemini 3.5 Flash Cyber?

Entroly does not advertise Gemini 3.5 Flash Cyber as a generally available API model. Google says CodeMender access with that model is restricted to selected governments and trusted partners.

Does listing a model guarantee token or dollar savings?

No. Compatibility means Entroly can apply its context contract on a supported route. Savings depend on the request, selected evidence, caching, output usage, and whether provider-bound traffic actually traverses Entroly.

Model compatibility · verified 2026-08-11

Verified model support, without announcement inflation.

Entroly selects and compresses context before a request reaches the model. It does not run hosted models itself. A verified listing requires a public model ID and official limits, or an inspectable local Ollama tag.

What stays the same across models

Budget-aware evidence selection, exact recovery handles, Context Receipts, optional verification, and conservative accounting. Provider tools, authentication, model execution, retention, and billing remain controlled by the configured provider or local runtime.

Current verified matrix

Model	Verified ID or tag	Entroly route	Limits used by the registry	Status
GPT-5.6 Sol	`gpt-5.6-sol`; alias `gpt-5.6`	OpenAI API proxy or supported agent path	1.05M context; 128K output	Verified
GPT-5.6 Terra	`gpt-5.6-terra`	OpenAI API proxy or supported agent path	1.05M context; 128K output	Verified
GPT-5.6 Luna	`gpt-5.6-luna`	OpenAI API proxy or supported agent path	1.05M context; 128K output	Verified
Gemini 3.6 Flash	`gemini-3.6-flash`	Gemini API proxy	1,048,576 input; 65,536 output	Verified
Gemini 3.5 Flash-Lite	`gemini-3.5-flash-lite`	Gemini API proxy	1,048,576 input; 65,536 output	Verified
NVIDIA Nemotron 3.5 Lightning	`nemotron-3.5-lightning` or installed variant	Loopback Ollama discovery + OpenAI-compatible proxy	Discovered from the installed tag	Verified at runtime

Gated or unverified announcements

Entroly does not manufacture model slugs or silently assign a large context window to a headline. Unknown models receive a visible conservative fallback until the registry or local discovery can establish stronger metadata.

Announcement	Why it is not in the verified public matrix
Gemini 3.5 Flash Cyber	Google says CodeMender access with this model is restricted to selected governments and trusted partners.
Meta Muse Spark	Meta describes the API as a private preview for selected users.
GPT-5.6 Cyber, Claude Opus 5, Muse Glimmer-30B, Muse Spark 1.2, Shieldstral 1.0	No official generally usable API model ID and complete public limits were verified for this release. Entroly therefore makes no exact-support claim yet.

Configure a supported route

# Gemini API path
set GOOGLE_GEMINI_BASE_URL=http://127.0.0.1:9377
entroly proxy

# Local Ollama path
python -m entroly.models discover ollama --inspect-ollama-context

Use export instead of set on macOS/Linux. A model appearing here does not guarantee savings; use Entroly receipts and provider usage to measure the actual workload.

Official sources

README · GPT-5.6 details · Nemotron/Ollama details