Last verified
VISUAL REASONINGIMAGE + VIDEO INPUTRESEARCH PREVIEWNO CACHE RATECONTEXT NOT PUBLISHED

Glimpse API Pricing

Glimpse is the first model from Elorian, a lab founded by former Google DeepMind researchers: a visual reasoning model that works on images and video directly - zooming into a region, rewatching a few seconds at a denser frame rate, tracking one feature across time - rather than turning the picture into text and reasoning about the text. Elorian prices it at $0.60/M input and $3.00/M output, with no cached-input rate. Released as a research preview on October 16, 2026, through Elorian's own API and chat platform. Checked by hand against Elorian's launch post.

Input - per 1M tokens
$0.60/M
Images and video billed as input launch rate
Output - per 1M tokens
$3.00/M
Reasoning billed as output launch rate
Cached input
n/a
No cache rate published N/A
Effective - agentic blend
$0.792/M
92/8 split - no cache discount
§ 01 / TERMINAL

Glimpse API cost calculator.

Live calculator pre-loaded with Glimpse's launch rates. Elorian publishes no cached-input rate, so every input token is priced in full. Images and video frames count as input tokens, and Elorian says the model looks closely and spends tokens doing it - size the output side generously until you have measured your own runs.

$ /mo
Workload split
Tokens you can process
—
Words equivalent (English)
—
Effective rate
—
Open full calculator (all models · share URL · CSV) →
§ 02 / SCENARIOS

Glimpse cost on real workloads.

CHARTS

Lab chart reasoning

$0.013/chart
6,000 in - 3,000 out~7,936 charts/$100
VIDEO

Short clip dense caption

$0.030/clip
30,000 in - 4,000 out~3,333 clips/$100
COUNTING

Product photo object count

$0.0063/photo
3,000 in - 1,500 out~15,873 photos/$100
ROBOTICS

Robot episode step check

$0.054/episode
60,000 in - 6,000 out~1,851 episodes/$100
§ 03 / TOKENIZER

Glimpse token counter and cost.

What's a token →
Estimate · glimpse-tokenizer-estimate · ≈3.85 chars/token Auto-counts as you type

This is a chars-per-token approximation, not a real tokenizer. Actual tokens vary by language, code density, and tool-call overhead — counts are typically ±10–20% off for English prose, more for code or non-Latin scripts. For exact billing, use the vendor's official tokenizer.

Characters 615
Words 106
Tokens (estimated) 160 tokens
Cost as input · uncached $0.0001 USD
Cost as output · uncached $0.00048 USD
§ 04 / SHELF

Glimpse vs similar models.

All models →
Model Input /M Output /M Effective blended Context Best for
Glimpse Current $0.60 $3.00 $0.792 agentic 92/8 not published Visual reasoning over images and video
Gemini 3.8 Flash $0.75 cache $0.075 $3.75 $0.481 −39% · promo 1M General multimodal model at a close rate
Perceptron Mk1.5 $0.15 cache $0.0375 $1.50 $0.173 −4.6× 36K Grounded image, video and audio perception
Gemini 3.5 Flash-Lite $0.30 cache $0.03 $2.50 $0.272 −2.9× 1M Budget general multimodal model
GPT-6.1 Sol $2.00 cache $0.10 $10.00 $1.29 +63% 1.05M Frontier model Elorian benchmarks against
Claude Opus 5.5 $4.00 cache $0.20 $20.00 $2.58 +3.3× 1M Anthropic's general frontier

Glimpse API pricing FAQ.

What Elorian Glimpse costs, what Elorian has not published yet, and how its rate sits next to the models it benchmarks against.

Q · 01 How much does Glimpse cost? +
Elorian's launch post gives $0.60/M input tokens (uncached) and $3.00/M output tokens. With no cached-input discount, our 92/8 agentic blend works out to $0.792/M.
Q · 02 Is there a cached-input rate or a batch discount? +
None is published. The launch post labels the input rate uncached without giving a cached one, and the pricing page on Elorian's platform sits behind a sign-in, so this page prices every input token in full. If Elorian publishes a cache or batch rate, it goes here with the date it was read.
Q · 03 What is the context window? +
Elorian has not published one, nor a maximum output length or an API model id outside its signed-in platform, so this page states none. Reseller listings carry their own numbers; we do not copy them as the vendor's.
Q · 04 What does Glimpse do differently? +
Elorian's claim is that it reasons in visual space instead of converting what it sees into words first: it revisits regions of an image, rewatches short stretches of video at a higher frame rate and follows one object through a moving shot before answering. It is also trained to use tools that run code, transform images and watch video, and to coordinate subagents. Elorian aims it at robotics, scientific data and video work.
Q · 05 How does its price compare with the models Elorian benchmarks against? +
Glimpse's $0.60/M input and $3.00/M output sit below Gemini 3.8 Flash at $0.75/M and $3.75/M on its introductory card, and far below GPT-6.1 Sol and Claude Opus 5.5. Elorian itself warns that close visual review takes time and tokens, so a lower rate per token does not guarantee a lower bill per task.
Q · 06 How good is it? +
Elorian's launch table, run by Elorian at the highest effort for every model, puts Glimpse ahead of GPT-6 Astra on four evaluations - chart reasoning, open-vocabulary counting, short-form video captioning and robot manipulation events - and level on embodied spatial reasoning, while it trails on others such as Video-MME v2 and dense video captioning. These are a vendor's numbers on a research preview; check them on your own images and clips.
Q · 07 How do I get access? +
Through Elorian's platform, which offers an API and a chat interface; sign-up is open, and enterprise volume access goes through Elorian's sales team. Glimpse is a research preview, so expect behaviour and terms to change. The weights are not released: Elorian says it post-trained an open-weights base model it does not name.
Y
Reviewed by Yaroslav Vikhariev Founder · AI//COST · Prices checked by hand against elorian.ai · Last verified Oct 19, 2026
Methodology Report a correction More by Y.V.