Skip to main content

image model comparison

GPT Image 2.6 Sunburst vs MAI Image 2.7

GPT Image 2.6 Sunburst and MAI Image 2.7 are both current image models, from OpenAI and Microsoft AI. On Arena Text-to-Image, GPT Image 2.6 Sunburst scores 1410 Elo and MAI Image 2.7 1387 Elo (Arena, measured November 17, 2026).

Higher Arena Text-to-Image
GPT Image 2.6 Sunburst1410 Elo against 1387 Elo
Newer release
MAI Image 2.7Nov 17, 2026 against Nov 4, 2026

Data verified November 18, 2026 · Dataset 2026-11-18

Share this

Side by side

GPT Image 2.6 Sunburst and MAI Image 2.7, line by line

Prices in effect today, compared at the same tier wherever both list one. The better value in each row is highlighted.

FactGPT Image 2.6 SunburstMAI Image 2.7
Price
Price per 1M input tokens$5Not listed
Independent benchmarks
Arena Text-to-Image1410 Elo (#2 of 12) (better)Range 1401 to 14191387 Elo (#4 of 12)Range 1375 to 1399
Arena Image EditToo close to call1481 Elo (#3 of 11)Range 1474 to 14881475 Elo (#4 of 11)Range 1464 to 1487
Specs
ProviderOpenAIMicrosoft AI
Release dateNovember 4, 2026November 17, 2026
AccessAPI onlyAPI only
Context windowNot reportedNot reported
Max outputNot reportedNot reported
Inputtext, imagetext, image
Outputimageimage

Ranges under the scores are the publishers' own: published rating ranges from Arena. Where two ranges overlap, the gap is inside the measurement's own uncertainty, so the row says too close to call and neither score is highlighted or counted as a lead.

Scores:Arena · 2026-11-18

Open both in the interactive comparison

Charts

Against every current image model

Where GPT Image 2.6 Sunburst and MAI Image 2.7 land among the current models of their task, on every independent benchmark, on price and on context.

Where each stands

On Arena Text-to-Image, GPT Image 2.6 Sunburst ranks #2 and MAI Image 2.7 #4 of 12 current models. Each bar is the share of current models it beats: on score, price and context.

GPT Image 2.6 Sunburst and MAI Image 2.7 among the current image models, one row per benchmark, price and context window.

Independent benchmarks

Arena Text-to-Image

GPT Image 2.6 Sunburst:#2of 12

1,410 Elo · ahead of 91%

Published rating range 1,401 to 1,419

MAI Image 2.7:#4of 12

1,387 Elo · ahead of 73%

Published rating range 1,375 to 1,399

Arena Image Edit

GPT Image 2.6 Sunburst:#3of 11

1,481 Elo · ahead of 80%

Published rating range 1,474 to 1,488

MAI Image 2.7:#4of 11

1,475 Elo · ahead of 70%

Published rating range 1,464 to 1,487

Cost

What each costs

Example workloads at each provider's listed rates, then every other rate either one lists. The cheaper figure in each row is highlighted.

Workload or rateGPT Image 2.6 SunburstMAI Image 2.7
Other listed rates
1M input tokens, cached$1.25Not listed
1M output tokens, batch image output$15Not listed
1M input tokens, image input$8Not listed
1M output tokens, image output$30Not listed

Your own bill will differ. These figures use list prices and example workloads. The AI Model Cost Audit reviews your real usage, shortlists models that score as well for each task for less, and estimates the savings.

Launch price $100, regular price $150

See the cost audit

Verdict

When to choose each

GPT Image 2.6 Sunburst

Where it leads

  • Higher Arena Text-to-Image score, 1410 Elo vs 1387 Elo
  • A cached-input rate, $1.25 per 1M tokens

Best for

  • Detailed compositions, brand-consistent edits and animated GIFs where quality matters most

Watch out for

  • Arena rates it below GPT Image 2.5 Sunburst: 1410 against 1428 for generation and 1481 against 1519 for editing (November 17, 2026), on 6,547 and 16,881 votes so far
  • Batch halves every rate, as for GPT Image 2.5: $4 per 1M image input and $15 per 1M image output tokens, for jobs that can wait

MAI Image 2.7

Where it leads

No measured lead over GPT Image 2.6 Sunburst in this data.

Best for

  • Edits that must keep the rest of a picture unchanged, such as swapping a product's color or material
  • Product images, logos and icons that need transparent backgrounds

Watch out for

  • Arena rates it below GPT Image 2.5 Sunburst on both boards (1428 and 1519) and below GPT Image 2.6 Sunburst for generation (1410), on 3,082 and 3,007 votes so far
  • Microsoft has published only an approximate price, about $35 per 1,000 images, and no rate per image or resolution, so no price is recorded here
  • A preview in Microsoft Foundry, under its preview terms

From benchmark to production

Building on GPT Image 2.6 Sunburst or MAI Image 2.7?

Public scores rarely hold on your own data. We test both models on your real tasks, then build the agent or pipeline around the one that wins.

AI Model Cost AuditLaunch price $100, regular price $150

FAQ

GPT Image 2.6 Sunburst vs MAI Image 2.7 FAQ

Do GPT Image 2.6 Sunburst and MAI Image 2.7 discount cached input?

GPT Image 2.6 Sunburst lists a cached-input rate of $1.25 per 1M tokens, 75% below its standard input rate. No cached-input rate is listed for MAI Image 2.7.

Which scores higher on independent benchmarks, GPT Image 2.6 Sunburst or MAI Image 2.7?

GPT Image 2.6 Sunburst leads on Arena Text-to-Image, 1410 Elo to 1387 Elo; GPT Image 2.6 Sunburst and MAI Image 2.7 are level on Arena Image Edit, 1481 Elo and 1475 Elo, inside overlapping published rating ranges (1474 to 1488 and 1464 to 1487). Scores are independent evaluations, not either vendor's launch figures.

AI Model Brief

The AI models worth knowing, once a week

What shipped, how it scores on independent benchmarks, what it costs per million tokens and what was deprecated, from the data behind our AI model benchmarks.

One email a week. We send a confirmation link first, and every issue has a one-click unsubscribe.

We use your address only to send the brief. See our privacy policy.

See this week’s brief

Production evaluation

Choosing between GPT Image 2.6 Sunburst and MAI Image 2.7?

Public benchmarks narrow the shortlist. We run both on your own tasks, data and cost limits, then build the routing and review controls to run the winner in production.

  • A head-to-head evaluation on your real workload
  • Routing between models by cost and difficulty