Skip to main content
Specialist modelPublic preview

Glimpse

Elorian

Released
October 16, 2026
Data date
October 24, 2026
Access
Elorian API and chat interface
Input price
$0.6
Per 1M input tokens

Glimpse is Elorian's agentic visual reasoning system for images and video. It can use tools and start subagents. In addition to Elorian's API and chat interface, OpenRouter documents it as elorian/glimpse. Its Together endpoint accepts text, images, and video, provides a 262,144-token context, and allows up to 256,000 output tokens within that context budget. These limits apply to the documented provider endpoint.

Specifications and access

SpecificationValue and source
Release
Research preview, announced on October 16, 2026Source
Access
Elorian API and chat interfaceSource
OpenRouter model ID
elorian/glimpse. The documented Together endpoint is elorian/glimpse-20261016.Source
Input modalities
Text, images, and video through OpenRouterSource
Context window
262,144 tokens through OpenRouter/TogetherSource
Maximum output
256,000 tokens through OpenRouter/Together, within the shared context budgetSource
Capabilities
Can use tools to write code, transform images, and inspect video; supports subagentsSource
Base model
Post-trained from an unnamed open-weights base; no Glimpse weight files or license specifiedSource
Announcement
2026-10-16Source
Product type
agentic visual reasoning systemSource
Input
images and video (The launch article describes visual inputs. OpenRouter separately documents text, image and video input for its Together endpoint.)Source
Public API model ID
not publishedSource
Base model
unnamed open-weights modelSource
Input price in USD per million tokens
0.6 tokensSource
Output price in USD per million tokens
3 tokensSource
Provider model ID
elorian/glimpse on OpenRouterSource
Deployment options
Together endpoint elorian/glimpse-20261016 through OpenRouterSource
Context window
262,144 tokens (Applies to the documented OpenRouter/Together endpoint, not an independently established direct Elorian API limit.)Source
Maximum output
256,000 tokens (Applies to the documented OpenRouter/Together endpoint, not an independently established direct Elorian API limit. Output shares the total context budget.)Source
API prices in USD per million tokens
USD 0.60 input; USD 3.00 output; USD 0.06 cache read per million tokens (These are OpenRouter endpoint tariffs; the cache rate is not established for Elorian direct billing.)Source
API capabilities
reasoning, include_reasoning, max_tokens, temperature, top_p, stop, frequency_penalty, presence_penalty, top_k, repetition_penalty, logit_bias, min_p, tools, tool_choice, response_format, structured_outputs, logprobs, top_logprobsSource
Input
Text, Image, VideoSource
Model code-generation failures
3Source
Provider infrastructure failures excluded
0Source
One-sided runtime faults
0Source
Requested output-token limit
235,929Source
Generation attempts in the index
23Source
Contributing programs with reasoning disabled
5Source
Contributing programs with reasoning enabled
8Source
Average estimated cost (USD)
0.08Source
Median estimated cost (USD)
0.05Source
Estimated cost range (USD)
$0.01-$0.22Source
Provider-reported total cost (USD)
1.31Source
DuelLab cost-data combinations
16Source
Average output token count
25kSource
Output-token range
2.1k-73kSource
Runs with output-token data
16Source
Player-program result distribution (labels not retained)
Reasoning enabled: 75% / 25% / 0%, n=8; reasoning disabled: 50% / 13% / 37%, n=8Source
GCB trust tier
automatedSource
GCB guardrail warning
Significant guardrail issues may impede work.Source
Busey run duration (seconds)
100Source
Busey run cost (USD)
0.0684Source
Busey total tokens
22,952 tokensSource
Reasoning tokens in output
9,636Source
Busey judge coverage
9/9Source
Busey judge count
3Source
Busey samples per judge
3Source
OpenRouter run ID
run-glimpse-openrouter-1792245634824-0fisgSource
Inference cost (USD, Build 1)
3.04Source
Total requests (Build 1)
299Source
Exact total tokens (Build 1)
20,158,149 tokensSource
Reasoning tokens (Build 1)
283,348 tokensSource
App build duration (Build 1)
37m 18sSource
Input tokens (rounded, Build 1)
19.8MSource
Output tokens (rounded, Build 1)
376kSource
Cached input tokens (rounded, Build 1)
18.4MSource
Glimpse model calls (Build 1)
298Source
Gemini helper calls (Build 1)
1Source
Inference cost (USD, Build 2)
2.42Source
Total requests (Build 2)
301Source
Exact total tokens (Build 2)
11,170,983 tokensSource
Reasoning tokens (Build 2)
209,961 tokensSource
App build duration (Build 2)
32m 19sSource
Input tokens (rounded, Build 2)
10.9MSource
Output tokens (rounded, Build 2)
285kSource
Cached input tokens (rounded, Build 2)
9.2MSource
Glimpse model calls (Build 2)
300Source
Gemini helper calls (Build 2)
1Source
Observed range, not a confidence interval
0.1595-0.1732Source
Comparison group
9Source