- OpenAI retired DALL-E from the API on May 12, 2026, and had already swapped it out of ChatGPT the year before; image generation there now runs on GPT Image, which reached version 2.6 on November 4, 2026, so “DALL-E” is no longer a live product to put on a 2026 shortlist.
- Google wound down Imagen 4 the same way — deprecated on Vertex AI in March 2026 and shut off the Gemini API entirely on August 17, 2026 — with image generation in the Gemini app and API now handled by the Nano Banana family, topped by Nano Banana Pro.
- Midjourney is still subscription-only with no permanent free tier, running $10 to $120 a month across four plans; V7 adds Omni Reference for holding a character or product consistent across a batch, as covered in our full Midjourney review.
- Adobe Firefly is the one tool here built around commercial-safety first: its enterprise plans come with contractual IP indemnification, because the underlying Firefly Image Model trains only on licensed and Adobe Stock content — see our Adobe Firefly review for the full breakdown.
| Five AI Image Generators at a Glance (November 2026) | |||||
|---|---|---|---|---|---|
| Midjourney | GPT Image (OpenAI) | Firefly (Adobe) | Nano Banana Pro (Google) | Flux 2 (Black Forest Labs) | |
| Current model | V7 (V8 in alpha) | GPT Image 2.6 | Firefly Image Model 5 | Gemini 3 Pro Image | Flux 2 [pro] / [max] / [klein] |
| Access | Discord / web app | ChatGPT, API | Firefly web app, Creative Cloud | Gemini app, API, Vertex AI | API via Black Forest Labs and resellers |
| Entry price | $10/mo (Basic) | Free limited in ChatGPT; Plus $20/mo for full use | Free (25 credits, watermarked); Standard $9.99/mo | Free limited in Gemini app; AI Pro $19.99/mo | From about $0.014/image (klein), no subscription |
| Commercial-use terms | Included on paid plans | Included on paid plans/API | Included, plus enterprise IP indemnification | Included on paid tiers; SynthID watermark always on | Varies by license tier; Dev weights are open but non-commercial |
The Lineup Looks Different Than It Did a Year Ago
Anyone comparing AI image generators by name alone will run into a problem fast: two of the five most commonly cited tools from a year ago don’t exist anymore under those names. OpenAI pulled DALL-E 2 and DALL-E 3 from its API on May 12, 2026, following an earlier switch inside ChatGPT itself, where DALL-E 3 had already been replaced by GPT Image’s native generation. The current release in that lineage, GPT Image 2.6, shipped November 4, 2026, with GPT Image 1 and several earlier variants already retired or on a published shutdown schedule. Google followed a similar path with Imagen: Imagen 3 and Imagen 4 endpoints were deprecated on Vertex AI in March 2026, and Imagen 4 access on the Gemini API closed on August 17, 2026. Image generation inside the Gemini app and API now runs on Google’s Nano Banana models, with Nano Banana Pro (built on Gemini 3 Pro Image) as the top tier. That leaves Midjourney, Adobe Firefly, and Black Forest Labs’ Flux as the names that have stayed stable, while the two biggest frontier-lab players rebranded and re-architected underneath familiar expectations.
Pricing Compared
None of the five tools price the same way, which makes a straight dollar comparison tricky. Midjourney is flat-rate subscription only — Basic at $10/month (roughly 200 generations), Standard at $30, Pro at $60 with stealth mode, and Mega at $120 for the heaviest individual users — with no permanent free tier. GPT Image is bundled into ChatGPT: a limited free allowance exists, but realistic daily use needs ChatGPT Plus at $20/month, or metered API access priced per image. Adobe Firefly has the lowest free entry point, 25 credits a month with a watermark, then Standard at $9.99, Pro at $19.99, Pro Plus at $49.99, and Premium at $199.99, with Creative Cloud subscribers also getting 1,000 Firefly credits included. Google’s Nano Banana Pro is available free with limits inside the Gemini app, or via the $19.99/month Google AI Pro plan for higher-volume use in-app; on the API, Nano Banana Pro runs about $0.134 per image at 1K/2K resolution and $0.24 at 4K. Flux breaks from subscriptions entirely — Black Forest Labs and resellers price per image by model tier and resolution, from roughly $0.014 for the small Flux 2 [klein] variant up to about $0.07 for Flux 2 [max], with Flux 2 Dev available as open weights for non-commercial local use.
| Tool | Cheapest paid access | Realistic daily-use tier | Per-image equivalent (approx.) |
|---|---|---|---|
| Midjourney | Basic, $10/mo | Standard, $30/mo | ~$0.03–$0.15 depending on plan/volume |
| GPT Image | ChatGPT free (limited) | ChatGPT Plus, $20/mo | Metered on API; bundled in ChatGPT |
| Adobe Firefly | Free, 25 credits/mo | Standard, $9.99/mo | Credit-based, varies by output type |
| Nano Banana Pro | Free (limited, in Gemini app) | Google AI Pro, $19.99/mo | $0.134 (1K/2K) – $0.24 (4K) on API |
| Flux 2 | No subscription; pay per image | Flux 2 [pro], ~$0.03/image | $0.014–$0.07 by tier/resolution |
Photorealism and Creative Control
Midjourney’s reputation for stylized, painterly, often strikingly art-directed output still holds in V7, which also added Omni Reference and expanded style references — uploading a reference image and having the model extract its palette, composition, and mood for a new generation. That makes it well suited to concept art, moodboards, and campaigns where a consistent aesthetic matters more than pixel-level editing control. GPT Image and Nano Banana Pro both lean toward instruction-following and targeted editing rather than painterly style: GPT Image can take image inputs and make localized edits without regenerating the whole frame, and Google markets Nano Banana Pro similarly around precise, prompt-faithful edits plus notably strong text rendering inside images. Firefly sits closer to a production tool than a creative-exploration one — it’s built to slot into Photoshop and other Creative Cloud apps for generative fill, background replacement, and product-scene work. Flux’s open architecture has made it a favorite for developers and technical teams who want to fine-tune or self-host a model rather than call a hosted API as-is; see our Flux review for how its open-weight and hosted tiers differ in practice.
Commercial Safety and Licensing
This is where the five tools diverge most sharply, and it matters more for commercial use than any visual quality difference. Adobe’s pitch is built entirely around this axis: Firefly’s models are trained on Adobe Stock, licensed content, and public domain material, and Adobe backs paid and enterprise plans with contractual IP indemnification — a formal commitment to defend customers against third-party copyright claims tied to Firefly output. None of the other four make that same indemnification commitment at the consumer-plan level. Google’s Nano Banana models apply SynthID watermarking to every generated image by default, which helps with provenance and disclosure but isn’t a licensing guarantee in the way Adobe’s indemnification is. Midjourney and GPT Image both grant commercial-use rights on paid tiers but without Adobe’s formal indemnification language, and Flux’s licensing varies by tier — the open Flux 2 Dev weights are explicitly non-commercial, while the hosted Pro, Flex, and Max tiers carry their own commercial terms through Black Forest Labs or a reseller. Teams generating imagery for paid ad campaigns, packaging, or anything facing legal review should read the specific license on the plan they’re paying for rather than assume parity across tools.
Speed and Workflow Fit
Where a tool lives changes how it gets used day to day. Firefly and Nano Banana Pro are both designed to sit inside a larger workflow — Firefly inside Creative Cloud apps, Nano Banana Pro inside the Gemini app, Google AI Studio, and Vertex AI — so they tend to get reached for mid-project, for a specific asset, rather than as a standalone creative destination. Midjourney’s Discord-and-web-app setup makes it more of a dedicated session: batch-generating variations, grading them, and iterating on a prompt chain. GPT Image’s home inside ChatGPT means it benefits from conversational back-and-forth — describing a change in plain language and getting a targeted edit back — which suits non-technical users more than a prompt-engineering-heavy workflow. Flux, accessed almost entirely through APIs and third-party platforms, is built for developers wiring image generation into an app or pipeline rather than for someone generating one image at a time in a browser tab, which is also why its pricing is metered per image instead of bundled into a subscription.
Who Each Tool Is For
Midjourney remains the strongest pick for art-directed, stylized visuals where mood and composition matter more than literal instruction-following — concept art, editorial illustration, campaign moodboards. Adobe Firefly is the right default for any commercial pipeline that needs defensible licensing, especially inside a Creative Cloud workflow already running Photoshop or Illustrator. GPT Image fits teams already inside the ChatGPT ecosystem who want conversational, targeted edits without learning a separate tool. Nano Banana Pro suits anyone standardized on Google’s stack — Workspace, Vertex AI, or the Gemini app — who wants strong text rendering and precise edits with built-in provenance watermarking; our guide to generating images in Gemini walks through that workflow end to end. Flux is the fit for developers and technical teams who want to self-host, fine-tune, or integrate image generation directly into a product rather than use it through a consumer-facing app.
Frequently Asked Questions
Is DALL-E still available in 2026? No. OpenAI retired DALL-E 3 from ChatGPT in 2025 and shut down the DALL-E 2 and DALL-E 3 API endpoints entirely on May 12, 2026. Anything currently generating images inside ChatGPT or the OpenAI API is running on GPT Image, not DALL-E.
What replaced Google’s Imagen 4? Imagen 4 was deprecated on Vertex AI in March 2026 and its Gemini API access closed on August 17, 2026. Google’s current image generation runs on the Nano Banana family, with Nano Banana Pro as the top-tier model available in the Gemini app, Google AI Studio, the Gemini API, and Vertex AI.
Which of these five is cheapest for occasional use? Adobe Firefly’s free tier (25 credits a month, watermarked) and Google’s free, limited Nano Banana access in the Gemini app are the lowest-cost starting points. For pay-as-you-go without any subscription, Flux 2’s smaller klein tier prices from roughly $0.014 per image.
Bottom Line
The five-way comparison that made sense a year ago doesn’t map cleanly onto November 2026: DALL-E and Imagen 4 are both retired, replaced by GPT Image and Nano Banana Pro respectively, while Midjourney, Adobe Firefly, and Flux have stayed put under their original names. The practical choice now comes down less to “which produces the best-looking image” and more to where the work happens and what the licensing needs to hold up to — Midjourney for art-directed style, Firefly for indemnified commercial use inside Creative Cloud, GPT Image for conversational editing inside ChatGPT, Nano Banana Pro for Google-stack workflows with built-in watermarking, and Flux for developers building image generation into their own products.