GPT Image 2.5
Last verified September 9, 2026 · OpenAI pricing ↗$5.00/1M input · $30.00/1M output · 32K context · OpenAI
Count Tokens
—
Tokens
—
Input Cost
—
Output Cost
Estimate Monthly Cost
Monthly Cost Estimator
Quick:
<$0.0001/mo
Pick a preset above or enter custom usage
Alternatives to GPT Image 2.5
Pricing Details
OpenAIImage generation and editing. Two snapshots released 2026-09-08, gpt-image-2.5-flare-2026-09-08 and gpt-image-2.5-sunburst-2026-09-08, on an IDENTICAL rate card: text input $5.00, cached text input $1.25, image input $8.00, cached image input $2.00, image output $30.00 per 1M. The stored output rate is the IMAGE output rate of $30.00/1M. Every line matches GPT Image 2 to the cent, which OpenAI states in the changelog ('use GPT Image 2 token rates') and on both model pages. NO BATCH RATE IS PUBLISHED for either snapshot: the pricing page's batch pane lists gpt-image-2 and the three retiring models but neither 2.5 model, though both list the v1/batch endpoint, so batch eligibility is unconfirmed. WHERE THE COST ACTUALLY MOVED: quality tiers were renumbered down one rung. At 1024x1024, tokens per image are low 196, medium 439, high 1,756, xhigh 3,122, max 7,024, giving $0.00588, $0.01317, $0.05268, $0.09366 and $0.21072, or $5.88 to $210.72 per 1,000 images. GPT Image 2 spent 196 / 1,756 / 7,024 at low / medium / high, so 2.5's high equals GPT Image 2's medium and 2.5's max equals its high, exactly. Callers who keep quality:'high' across the model swap get a 75% cut; callers who move to max pay exactly what they paid before. Ceiling unchanged at $0.21072 and both default to quality:'auto'. xhigh and max have no GPT Image 2 equivalent. OpenAI publishes a static tokens-per-image table only for models PRIOR TO gpt-image-2 (272/1,056/4,160 at 1024x1024); for these models it ships only an interactive calculator, whose constants (16/24/48/64/96, tokens = ceil(m^2 * aspect * (2e6 + W*H) / 4e6)) reproduce OpenAI's published GPT Image 2 dollar table on all three square cells. Non-square is cheaper at every tier: 1024x1536 costs less than 1024x1024. STREAMING: each partial image adds 100 image output tokens. SUNBURST VS FLARE: identical prices, rate limits (T1 100K TPM / 5 IPM to T5 8M TPM / 250 IPM) and endpoint support; badges differ (Sunburst highest performance/medium speed, Flare higher/very fast). OpenAI's guide states the two 'can use different token counts for the same quality setting and share the same price per image output token', so per-image cost can differ between them, but its calculator offers one combined option covering both, so no published figure distinguishes them. Not yet scored by Artificial Analysis, whose image leaderboard still ranks GPT Image 2 (high) first. Not present in LiteLLM's shared price file as of 2026-09-09. Model IDs: gpt-image-2.5-flare, gpt-image-2.5-sunburst.
Input / 1M tokens
$5
Output / 1M tokens
$30
Context Window
32K
Max Output
7K
Price History
Launched at current price on 2026-09-08. No price changes recorded since.
Frequently Asked Questions
Common questions about GPT Image 2.5 pricing and usage