Why must image tokens be accounted for in the context budget separately from text tokens?

Attempt it

Images are converted to tokens too, but the per-image cost is a function of resolution and provider-specific tiling rules, a high-resolution image can be 1,000 or more tokens. A budget calculator that only counts text will quietly under-report the real prompt length and start hitting overflow or hidden truncation. Treat images as a separate accounting line: compute tokens per image from the provider's published formula, add it to the running total, and apply the same eviction rules (drop the least-relevant image first under budget pressure).

Your answer

Multimodal models convert images into tokens before they can attend to them. Those **image tokens count against the same context window** as text tokens, they are not free. The number of tokens per image depends on **resolution and provider-specific tiling rules**. - A small thumbnail (under 512 pixels on each side) is typically 100-300 tokens. - A medium image (1024x1024) is typically 500-1000 tokens. - A high-resolution image (2048x2048 or larger) can be 1500-3000+ tokens. For reference, 1500 tokens is roughly equivalent to **2-3 pages of dense text**. One high-res image can equal several pages. If your budget calculator only counts text, it will **under-report the real prompt length**. The model receives a prompt longer than your calculator thinks, which leads to one of three bad outcomes: 1. The prompt overflows the hard window limit and the API returns an error. 2. The provider truncates silently (some APIs do this by default) and the model sees a clipped prompt without any indication. 3. The prompt fits but is far closer to the soft quality budget than expected, and answer quality degrades. The fix is straightforward. Treat images as a **separate accounting line** in the budget. For each image, compute its token cost using the provider's published formula (Anthropic, OpenAI, and Google all publish their image-token math). Add it to the running total. Apply the same eviction rules under budget pressure: drop the least-relevant image first. The rule of thumb: **assume an image is at least 1k tokens until you have measured otherwise**. That gives you a safe upper bound for budget planning.

**Images consume real tokens in the context window**, computed by provider-specific rules that depend on resolution and tiling. A text-only budget calculator silently under-reports prompt length and creates three failure modes: hard-limit overflow with an API error, silent truncation that drops content without notification, or soft-budget quality degradation that the engineer cannot explain. **The token math varies by provider.** **Anthropic's Claude Opus 4.7** computes image tokens as approximately `tokens = (width * height) / 750`, with a maximum of about 1600 tokens per image. A 1568x1568 image is roughly 1600 tokens; a 512x512 image is roughly 350 tokens. **OpenAI's GPT-5.5** uses a fixed cost per tile scheme: a baseline of about 85 tokens plus 170 tokens per 512x512 tile in detail-high mode. A 2048x2048 image is about 16 tiles plus baseline, around 2800 tokens. **Google's Gemini 3.1 Pro** uses approximately 258 tokens per image for typical resolutions, with adjustments for very large images. The exact numbers differ enough that **you cannot guess across providers**, you have to know the math for the model you are using. **Why this breaks a text-only budget.** If your calculator counts text tokens and assumes images are negligible, a prompt that the calculator says is 50k tokens may actually be 60k or 70k tokens depending on the images. When the hard limit is 200k, you might still fit. When the hard limit is 128k, you may overflow. When you are operating near a soft quality budget, you have shifted from comfortably-under to over-budget without realizing it. The failure modes: 1. **Hard overflow**: the API returns a context-length error. Easy to detect, hard to recover from in production. 2. **Silent truncation**: some APIs truncate the prompt to fit and proceed. The model produces an answer that ignores the truncated content. The engineer sees a confusing or wrong answer with no obvious cause. 3.

Quick reference

TL;DR

Images consume real tokens via provider-specific tiling rules, often 1000+ for a high-resolution image, so a text-only budget calculator under-reports prompt length and triggers silent overflow or truncation.

Memory aid

Sign in for the mnemonic that makes this stick. See Pro pricing.

Easy to grasp

Imagine a suitcase that has a weight limit. You weigh all your clothes carefully and pack everything to the line. Then you remember you also packed a heavy book, but you never weighed it. The suitcase is now over the limit and the airline starts removing things at the gate, and you do not know which. Images in a prompt are like that book. They take up real space in the context window even though they are not text. If your budget only weighs text, you go over without realizing it, and the model silently drops content.

Key concepts

Context Engineering

Detailed answer & concept explanation

~6 min read

Everything you need to truly understand this topic: intuition, mechanics, step by step explanation, code, formulas, and worked example. Click to expand.

Everything you need to truly understand this topic: intuition, mechanics, step by step explanation, code, formulas, and worked example.

Sign in for 5 free unlocks/day or go Pro for unlimited. See Pro pricing.

Quick scan · 2–4 min

Everything important, quickly.

Explain that images convert to tokens computed by provider-specific formulas dependent on resolution and tiling. Give an order of magnitude figure (1k+ tokens for a high-res image). Describe the three failure modes: hard overflow, silent truncation, and soft-budget degradation. Recommend treating images as a separate accounting line with provider-specific token-counting and the same eviction policies as text. Cover downscaling as a control, high-detail versus low-detail modes, and the instrumentation needed to detect silent truncation in production.

Real-world examples

Real products, models, and research that use this idea.

Anthropic's vision documentation publishes the exact per-image token formula for Claude Opus 4.7
OpenAI's vision documentation publishes the tile-based formula for GPT-5.5 with detail-high and detail-low modes
Google's Gemini documentation publishes per-image token costs for Gemini 3.1 Pro
Production document-AI products like AWS Bedrock applications track image tokens separately in their cost dashboards
Vision-RAG products in 2026 routinely downscale ingested images to 1024x1024 to balance quality and token cost

Interviewer follow-ups

What an interviewer would ask next. Try answering before peeking at the approach.

QHow would you decide between downscaling an image and sending it at full resolution?

Sign in or go Pro to see the approach for this follow-up.

QWhat instrumentation would you add to detect silent truncation in production?

Sign in or go Pro to see the approach for this follow-up.

QWhy does the right default in 2026 lean toward low-detail mode where providers offer it?

Sign in or go Pro to see the approach for this follow-up.

Don't say this

Red flags and common mistakes that signal junior thinking. Click to expand.

Most common mistake

Building a token-budget calculator that only counts text tokens, then watching the model truncate or behave strangely after a high-resolution image is added to the prompt.

Sign in or go Pro to see all red flags and common mistakes. See Pro pricing.

60-second cheat sheet

The night-before-the-interview bullets. Scan these on the way to the call.

Sign in for 5 unlocks/day or go Pro for unlimited revision sheets. See Pro pricing.

References

Primary sources. Skim if you want the original framing.