Elorian Glimpse

Glimpse is Elorian’s image and video reasoning model, announced as a research preview on October 16, 2026. See what makes this AI model special!

Glimpse explained
Release context, visual-thinking examples and evaluation caveats.
OpenRouter hosted offer
Check the selected provider’s current limits and access before sending a request.
Vercel AI Gateway offer
Use the separate Gateway contract, including its smaller output ceiling.

Visual ReasoningImage UnderstandingVideo UnderstandingElorian

Hosted API; Together also lists dedicated deployment. Self-hosting requirements are unverified.

USD per 1M tokens: $0.60 input / $3 output; OpenRouter and AI Gateway list $0.06 cache read. Rates and cache terms are host-specific; visual inspection and reasoning affect total token use.

Scoped Terms

Visual examples

Official Elorian platform recording of Glimpse inspecting and counting construction vehicles. · 36.767 seconds · No audio

Elorian records its own chat platform inspecting a construction-site video and returning a count with timestamps. This demonstrates the publisher platform; it does not establish that a reseller API exposes the same scratch space or tools. No independent API run is shown.

Elorian · original recorded platform session · Open video file

🚀Function Overview

Hosted visual understanding with text answers. Ready to experience the power of AI? Start your journey here!

Key Features

OpenRouter lists text, image and video input with text output.
OpenRouter’s Together endpoint: 262,144 context tokens; 256,000 maximum completion tokens, as advertised by that provider.
Vercel AI Gateway: 220,000 context tokens; 4,096 maximum output tokens, including reasoning tokens.
OpenRouter currently lists tools and structured_outputs among supported parameters; support in that host does not establish the full Elorian chat agent.
Together’s own model card lists dedicated deployment and deployment-specific endpoint details.
Official evaluation scores use publisher-selected harnesses and thinking settings. No universal quality ranking or native video output is established.

Use Cases

•Review chart or image details
•Describe video events and count visible objects
•Evaluate visual task accuracy against labeled references

⚙️Input Parameters

model

string

OpenRouter Chat Completions: elorian/glimpse at https://openrouter.ai/api/v1/chat/completions. This is a host ID, not a verified native Elorian API ID.

messages

array

OpenRouter role/content messages. Image parts use type=image_url with image_url.url. Use the host’s video-input documentation for videos; the text example below does not submit a video.

max_tokens

integer

OpenRouter completion cap within the selected provider’s budget; set an explicit budget. Vercel AI SDK uses maxOutputTokens instead, with its separate 4,096-token ceiling.

💡Usage Examples

Example 1

Input Parameters

{
  "model": "elorian/glimpse",
  "messages": [
    {
      "role": "user",
      "content": "List three visual checks needed before deciding whether a train in a clip is slowing down."
    }
  ],
  "max_tokens": 512
}

Output Results

Documented OpenRouter request shape, not executed here. This text-only example sends no clip and cannot establish visual performance. Inspect choices[].message.content and usage in the actual response.

Quick Actions

Technical Specifications

Hardware Type: Hosted API; Together also lists dedicated deployment. Self-hosting requirements are unverified.
Commercial Use: Hosted service use is governed by the selected provider and Elorian terms. No downloadable-weight or asset-reuse license is established. Review data handling and your upload rights before use.
Pricing: USD per 1M tokens: $0.60 input / $3 output; OpenRouter and AI Gateway list $0.06 cache read. Rates and cache terms are host-specific; visual inspection and reasoning affect total token use.