AI Image Editing Cost Calculator (USD + LKR)

Editing endpoints bill differently from plain text-to-image. When you call an edit endpoint you upload the image you want changed, and that input image is billed on top of the prompt and the generated output. Editing endpoints bill for the input image(s) you upload plus the prompt plus the output image. A plain text-to-image generation price only counts the output, so it under-states an edit — on token-metered models by the input-image and prompt tokens, which this tool breaks out below.The calculator models each provider's real billing scheme rather than a single flat rate.

There are four schemes among the six models:

• Gemini 2.5 Flash Image (Nano Banana) — flat output of 1,290 tokens × $30/1M = $0.039 per image, plus text and input images at $0.30/1M (an input image ≈ 258 tokens).
• GPT Image 1.5 — fully token-metered: text input $5/1M, image input $8/1M, image output $32/1M. Output tokens depend on quality and size (272 / 1,056 / 4,160 for low / medium / high at 1024×1024); input images use 32-pixel patches capped at 1,536.
• FLUX.1 Kontext [pro] / [max] — flat $0.04 / $0.08 per edited image; the input image is included.
• FLUX.2 [pro] and Qwen-Image-Edit — per output megapixel ($0.045/MP and $0.03/MP). A 1024×1024 image is 1.049 MP; a 1024×1536 image is 1.573 MP.

The math for a single edit is:

prompt_tokens       = round(prompt_words / 0.75)

# token-metered (GPT Image 1.5)
edit_usd = prompt_tokens        × 5  / 1e6
         + input_img_tokens     × 8  / 1e6
         + output_img_tokens    × 32 / 1e6

# Nano Banana
edit_usd = (input_img_tokens + prompt_tokens) × 0.30 / 1e6
         + 1290 × 30 / 1e6

# flat (FLUX Kontext)        edit_usd = flat_price
# per-megapixel (FLUX.2, Qwen) edit_usd = rate_per_mp × megapixels

batch_usd = edit_usd × number_of_edits
batch_lkr = batch_usd × usd_to_lkr_rate

Nothing is sent to a server — the whole pricing table is bundled into the page and all math runs locally, so your batch sizes and prompts stay private.

Pricing cross-check (computed live on this page) — each per-image number is re-derived from raw tokens and matched to the vendor's published figure:

• ✓Nano Banana output image = 1290 × $30/1M vs Google's $0.039/image → expected $0.039, got $0.0387
• ✓GPT Image 1.5 high 1024² output = 4160 × $32/1M vs OpenAI's $0.133/image → expected $0.133, got $0.1331
• ✓GPT Image 1.5 medium 1024² output = 1056 × $32/1M vs OpenAI's $0.034/image → expected $0.034, got $0.0338
• ✓GPT Image 1.5 low 1024² output = 272 × $32/1M vs OpenAI's $0.009/image → expected $0.009, got $0.0087

Worked-example self-test (also computed live) — each row reconciles the formula above with the hand-derived batch totals in the code header:

• ✓Nano Banana · 300 edits · 1 input · 30 words · 1024² → expected $11.64, got $11.64
• ✓GPT Image 1.5 · 50 edits · high · 1024² · 1 input · 30 words → expected $7.08, got $7.08
• ✓FLUX.1 Kontext [pro] · 50 edits · flat → expected $2.00, got $2.00
• ✓Qwen-Image-Edit · 100 edits · portrait (per-MP) → expected $4.72, got $4.72
• ✓Edge · 0 edits · Nano Banana → expected $0.00, got $0.00
• ✓Large batch · 100,000 edits · Nano Banana → expected $3878.94, got $3878.94

• OpenAI — API pricing (GPT Image 1.5 text / image input / image output rates)
• OpenAI — GPT Image 1.5 model card (per-image output token counts)
• Google — Gemini API pricing (Gemini 2.5 Flash Image / Nano Banana)
• Black Forest Labs — FLUX API pricing (FLUX.2 [pro] editing, per megapixel)
• Black Forest Labs — Documentation pricing (FLUX.1 Kontext [pro] / [max])
• fal.ai — Qwen-Image-Edit model pricing (per megapixel)
• Central Bank of Sri Lanka — Daily indicative exchange rates (USD→LKR)

Every rate was cross-checked against the vendor's live pricing page on 2026-07-17. Image pricing changes often; if you spot a stale row, email the address below — most fixes ship within 24 hours.