AI Image Generation Market Intelligence Report — September 2026

Key takeaways

• September recorded 192.46 million estimated visits across 227 image-generation-classified domains. Comparable monthly traffic fell 3.02%, while estimated visits per calendar day rose 0.22%. Daily intensity was broadly flat; the monthly decline alone overstates the immediate slowdown.
• The longer view remains weaker: 176 domains observed throughout May–September fell 13.84%. Excluding Magnific, the same fixed cohort fell 4.84%. This is an endpoint change across five monthly observations, not a monthly growth rate.
• The top ten domains capture 84.35% of September visits; Magnific alone captures 41.52%. These are shares of the tracked domain basket, not shares of generated images or revenue.
• Adobe Firefly supplied 51.31% of gross gains. Across 185 comparable domains, 4.67 million gross gains were outweighed by 10.66 million gross losses. There were 86 gainers, 98 decliners and one unchanged domain.
• Style-focused products retained 18.79% growth against June despite a 0.62% monthly dip. People and Portraits grew 3.54%, but fell 9.39% excluding PFPMaker. Neither signal establishes stronger monetization or retention.

01 Market Pulse

Monthly volume softened, but the calendar-adjusted reading is almost unchanged.

The comparable basket declined from 198.45 million visits in August to 192.46 million in September. September has 30 days versus 31 in August: daily-normalized visits actually increased 0.22%. This arithmetic adjustment does not control for weekday mix, seasonality or estimation error.

The medium-term picture is weaker. A fixed set of 176 domains declined at every monthly observation after May, reaching 192.30 million visits in September versus 223.18 million in May. Keeping membership fixed avoids confusing changes in directory coverage with changes in traffic, but cannot correct provider revisions or changes in what a domain contains.

02 Market Structure

Ten domains capture 84.35% of visits, led by Magnific at 41.52%.

Magnific contributes 79.92 million visits, Adobe Firefly 23.01 million and Picsart 15.91 million. The remaining 217 domains outside the top ten account for 15.65%. A small number of broad creative platforms therefore shape the category headline.

The Magnific listing points to an image-generator path, but the observation covers magnific.com as stored by the provider. Whole-domain traffic can include other activities. These figures should not be interpreted as shares of generated images, paying customers or underlying model use.

03 Segment Performance

Style-focused products hold their June gains; the portrait rebound depends on one domain.

General Image Generation records 161.23 million visits and declines 3.58% month over month and 12.44% against June. Style and Subject-Specific Images records 26.32 million visits: a 0.62% monthly dip, but 18.79% growth against June. People and Portraits records 4.92 million visits, rising 3.54% monthly while remaining 5.03% below June.

04 Growth and Loss

A few large gains could not offset a larger pool of losses.

Gross gains of 4.67 million visits were offset by gross losses of 10.66 million, leaving a net decline of 5.99 million. Firefly added 2.40 million visits. With PFPMaker, Krea, NovelAI and Moescape, the five largest gainers supplied 79.91% of gross gains.

Magnific and Picsart lost 2.70 million and 2.04 million visits respectively; Leonardo AI and Midjourney also declined. Breadth was modestly negative: 46.49% of comparable domains grew, 52.97% declined and 0.54% were flat. The largest movements determine much of the aggregate result.

05 Companies to Watch

Five cases illustrate growth, resilience and measurement risk.

Event context: Freepik acquired Magnific in May 2024 and announced the Freepik-to-Magnific rebrand on April 28, 2026. The latter date is relevant to the observed discontinuity. It does not establish how much of the jump reflects redirects, changed coverage or new activity. Sources: https://www.magnific.com/blog/freepik-is-now-magnific/ ; https://www.magnific.com/blog/from-freepik-to-magnific-the-story-of-europes-largest-ai-creative-platform/

06 Emerging Demand Signals

Channel patterns are leads for research, not verified explanations of demand.

PFPMaker combines rapid growth with a high organic-search share; Firefly combines growth with a rising organic-search share and substantial referrals. These observations justify looking at queries and referring domains. They do not yet identify which customer tasks became more popular.

PixAI, NovelAI and Moescape have September direct shares of 65.13%, 74.23% and 75.37%. This suggests an audience hypothesis worth testing, but direct traffic includes bookmarks and unattributed visits. Multi-purpose products also attract activity beyond image generation.

07 Opportunity Signals

Prioritize focused customer research over broad market-growth claims.

08 What Founders Can Learn

Combine comparable measurement with customer and product evidence.

Observation: monthly visits fell 3.02% while daily-normalized visits rose 0.22%. Implication: compare equivalent periods before treating a traffic change as a product problem. Confidence: medium; day normalization does not remove seasonality. Watch next: October daily visits and returning-user cohorts.

Observation: Firefly supplies more than half of gross gains, and PFPMaker reverses the portrait segment result. Implication: examine contribution and breadth before turning one winner into a market thesis. Confidence: medium; arithmetic is clear but business drivers are not. Watch next: results excluding the largest gainer.

Observation: broad creative domains dominate estimated visits. Implication: evaluate a proposed workflow with activation, repeated task completion and conversion alongside acquisition. Confidence: medium as a measurement implication; these data do not establish which business model will succeed. Watch next: feature-specific funnels and interviews.

09 What to Watch Next

Track whether October growth broadens beyond the largest contributors using the same cohort definitions.

Methodology and sources

Scope: the current Sigpik Image Generation primary-category route (taxonomy term 4), including General Image Generation, People and Portraits, and Style and Subject-Specific Images. Image Processing is a separate root and is excluded. Multi-purpose tools are included only when their selected primary leaf is in scope. This is a tracked basket, not a census of the industry.

Eligibility follows the existing route query: published tools and English localization; safe status; no duplicate_of_tool_id; verified or pending verification; fresh or aging staleness; an active primary leaf assignment that is verified or auto-accepted with confidence at least 0.5. Existing assignment precedence is preserved. Every normalized domain appears once. Current taxonomy membership is applied retrospectively to history; this is not a reconstruction of the historical directory.

Extracted on 8 October 2026 from Similarweb-source domain_traffic_monthly records, metrics schema 2. Only nonnegative integer visit observations are accepted. The query returned 1,266 records, below its 6,000-row cap. All 227 members have a valid September observation. Recorded zero estimates are retained; a zero does not establish an absence of real activity.

September total visits are 192,462,034. MoM is (192,458,281 / 198,446,095 − 1) × 100 = −3.02%, using 185 domains with valid September observations and positive August baselines. The other 42 domains have recorded zero August baselines and contribute 3,753 September visits. They enter current totals and shares but not MoM, breadth or gain/loss. They are not automatically classified as new launches.

Breadth: 86 gainers (46.49%), 98 decliners (52.97%) and one unchanged domain (0.54%), all within the 185-domain comparable cohort. Daily normalization uses 30 September days versus 31 August days: (192,458,281 / 30) / (198,446,095 / 31) − 1 = +0.22%. It does not control for weekday mix, seasonality or uncertainty in the provider estimates.

Fixed cohort: 176 domains with every May–September observation and positive May visits. Traffic falls from 223,180,881 to 192,295,234 (−13.84%). Five monthly observations span four month-to-month intervals. April remains in the raw snapshot but is excluded from the main trend baseline because Magnific changes from 9,418,112 to 105,081,872 between April and May around the documented April 28 rebrand. The extract does not quantify rebranding, redirect or provider-mapping effects.

Sensitivity checks: September MoM excluding Magnific is -2.84%, and excluding Firefly is -4.72%. May–September fixed-cohort change excluding Magnific is -4.84%. People and Portraits MoM excluding PFPMaker is -9.39%. These are descriptive robustness checks, not alternative market definitions.

Segment MoM uses positive-August cohorts: 118 general, 35 people and 32 style domains. Three-month change means September versus June, using valid September observations and positive June baselines: 117 general, 36 people and 32 style domains. It is not CMGR or CAGR. Segment counts and September traffic reconcile to the market because primary-category assignments are mutually exclusive.

Market shares use all 227 September observations. Share changes are percentage-point differences within the 185-domain comparable basket and therefore use a slightly different denominator. CR10 and segment CR3 sum the largest ten and three domain shares. They do not measure company ownership or unique audiences.

Gross gains sum positive September-minus-August deltas; gross losses sum signed negative deltas. 4,672,704 + (−10,660,518) = −5,987,814. Growth contribution uses gross gains, not net change, as denominator. The top five gainers contribute 79.91%.

Channel shares are frozen provider fractions displayed as percentages. Search includes organic and paid components; social includes organic and paid components. Do not add parent and child fields together. Direct traffic is not retention, and channel mix does not identify campaigns, queries or incremental acquisition.

Domain traffic includes other product activities. Magnific is recorded as magnific.com with the existing freepik-magnific slug; PFPMaker is pfpmaker.com under the existing snapheadshots slug. Identifiers are preserved; analytical references are the frozen domain and tool ID. Historical domain continuity and the PFPMaker profile route remain unverified; this limits historical and entity-level interpretation. A legacy slug is not evidence of an additional product.

Confidence refers to interpretation, not provider accuracy. Official company and product documentation adds context, not numerical attribution. No independent revenue, conversion, retention, keyword, campaign or country-baseline data were joined. Geographic Signals is omitted because no frozen country baseline is available. Watch thresholds are editorial monitoring rules, not statistical tests or forecasts.

The query, observations, Research JSON and snapshot hash are saved together. Figures are frozen; live product links may display later data. Historical comparisons remain subject to the domain-continuity limitations described above.

Off-web coverage: this basket does not directly count API calls, desktop or plugin generation, local model runs, Discord bot actions or native-mobile generation. Some associated website visits may be counted, but generation volume is not. Migration toward these channels is a low-confidence hypothesis, not an observed cause of the decline. Adobe documentation confirms APIs and embedded Photoshop generation exist; it does not quantify migration.

Seasonal attribution: back-to-school, quarter-end work and conference marketing are candidate checks. The report does not establish a recurring September demand uplift or a September Firefly campaign. Adobe MAX 2026 is scheduled for November 10–12; event timing alone cannot explain September traffic.

Display policy: prose and watch metrics use two decimal places with explicit percentage units. Machine-readable metrics and the frozen snapshot retain full precision for reconciliation. External context was checked on 8 October 2026; no traffic values were changed.

Sigpik Image Generation market; live taxonomy reference

Similarweb; source estimates are frozen within the report snapshot

Magnific: Freepik becomes Magnific (official rebrand announcement) · 2026-04-28

Magnific: company history and May 2024 acquisition · 2026-06-26

Adobe MAX 2026 official FAQ; schedule accessed 8 October 2026

Adobe Firefly Services: API documentation; accessed 8 October 2026

Adobe: Generative AI features in Photoshop; updated 20 May 2026