Inside the World’s Largest AI Film Festival: 10 Things to Know About Astana 2026

By the Filmoria News Desk

Astana, host city of the 2026 Astana AI Film Festival. Photo: Ken and Nyetta, via Wikimedia Commons (CC BY 2.0).

The Astana AI Film Festival is a competition for short films made primarily with generative AI tools, and with 8,067 entries from 125 countries it is now the largest event of its kind anywhere in the world.

That sentence would have sounded absurd two years ago. AI film competitions were a novelty slot, a few hundred clips judged by the people who made the software. This one has a $2 million fund behind it, a national technology cluster running it, ByteDance’s enterprise arm as a technology partner, and more submitted footage than most countries produce in a year. It takes place in Astana from 1 to 3 October 2026, and the awards are handed out on the final night.

Whether any of the films are any good is a separate question, and one nobody can answer until the finalists screen. What is already measurable is the scale, the money and who turned up — including 324 entries from the United Kingdom, which is more than Japan, Brazil or France managed.

How this list is ordered: scale first, then the money, then the people and rules behind it, then what it means for British filmmakers, and finally how it stacks up against every other AI film prize. Figures come from the festival’s own published statistics and from Kazakh and international press reporting; where the two disagree, both numbers are given.

1. 8,067 Films From 125 Countries (Astana, 2026)

What makes it stand out: no AI film event has ever reported a submission pile anywhere near this size.

Key data: 8,067 entries; 125 countries; 4,929 films in the open competition and 3,138 in the thematic strand; submissions open 25 May to 7 September 2026; entry free.

Why it matters: it settles the argument about whether generative filmmaking has an audience of makers. It plainly does.

For context, the Sundance Film Festival receives somewhere north of 4,000 feature submissions in a typical year, and that is from an industry with a century of infrastructure behind it. Astana got double that in shorts, in its first edition, from people who mostly did not exist as filmmakers eighteen months ago. The festival describes more than 8,000 filmmakers involved; The Astana Times reports 4,756 unique creators, which suggests a lot of teams and a lot of repeat entrants. Either way, the number that matters is 8,067 finished pieces of work, not 8,067 people saying they intend to make something.

Board chairman and co-founder Almaz Zhali framed the shift like this: “These are no longer isolated AI experiments — we are seeing full-fledged films, complex stories, characters and entire universes.” He would say that, running the festival. But the volume does back him up.

Photo: Matti Blume, via Wikimedia Commons (CC BY-SA 4.0).

2. $2 Million, and Only Half of It Is Prize Money

What makes it stand out: the fund is split between cash prizes and an investment pot meant to keep winning projects alive after the festival ends.

Key data: $2,000,000 total; $1,000,000 in competition prizes; $1,000,000 production and investment fund, provided by Balaji Srinivasan and selected through his Network School.

Why it matters: most festival prizes are a cheque and a photograph. This one is trying to function as a development slate.

The distinction gets lost in a lot of the coverage, which reports the fund as either $1 million or $2 million depending on which half the writer found first. Both are correct. The competition prizes total $1 million. Alongside them sits a separate $1 million of production capital, and the selections for that are made personally by Srinivasan and the Network School rather than by the festival jury.

Co-founder Aizatulla Khussain explained the reasoning: “We want strong projects to continue after the festival. That is why, in addition to the prize fund, there is a separate investment capital.” Finalists are also offered a route into Astana Hub itself, which is the part that looks least like a film festival and most like a startup accelerator with a screening room attached.

3. “The Future Worth Living In” Takes the Biggest Cheque

What makes it stand out: the thematic strand carries three quarters of the prize money and a single brief.

Key data: $750,000 total; grand prize $450,000; second $200,000; third $100,000; 3,138 entries.

Why it matters: $450,000 for a short film of up to ten minutes is, as far as we can tell, the largest single prize ever offered for a short in any format, AI or otherwise.

The theme is doing real work here. “The Future Worth Living In” is an optimism brief, which is an interesting instruction to hand a technology whose cultural reputation is currently somewhere between anxious and hostile. It rules out the easy dystopia that generative video defaults to — the ash-grey cities, the drone swarms, the collapsing coastlines — and asks for something harder to render convincingly: ordinary life, working well.

Whether the entries deliver that or just produce 3,138 variations on gleaming solarpunk skylines is one of the few genuinely interesting critical questions the festival raises. We will know when the ten thematic finalists screen.

Photo: Projjal Moitra, via Wikimedia Commons (CC BY-SA 4.0).

4. The Open Competition Pays Five Different Kinds of Good

What makes it stand out: instead of one best-film award, the open section splits its money across five craft categories worth $50,000 each.

Key data: $250,000 total; Best Character, Best Direction, Best Visual, Best Story and Best Concept; 4,929 entries across 19 genres; 15 finalists.

Why it matters: the category list is an argument about what generative filmmaking is currently bad at.

Look at what is being rewarded. Best Character, in a medium notorious for faces that change between shots. Best Story, in a medium that produces spectacular eight-second fragments and struggles to sustain a second act. Best Direction, in a medium where “direction” often means prompt iteration rather than working with people. The festival has effectively priced the field’s weak points at $50,000 apiece and invited entrants to prove them wrong.

Best Visual is the one that should be easiest. It is also, probably, the one where the standard will be highest, because that is where the tools are strongest and the competition thickest.

Nineteen genres is the other detail worth noting. Generative video has a house style — slow push-ins, shallow focus, a slightly damp cinematic grade — and genre is the thing most likely to break it. Comedy needs timing the tools cannot yet hold. Horror needs restraint, and these models are incapable of restraint unless told firmly. Documentary is close to impossible when nothing on screen happened. If the open competition produces one genuinely funny five-minute film, that will be a more significant result than anything in the thematic strand.

5. America Entered the Most Films. Britain Came Sixth.

What makes it stand out: the country table is not the one you would predict from where the AI industry sits.

Key data: United States around 1,395 entries; Kazakhstan 829; India roughly 700; China between 479 and 550 depending on source; South Korea about 350; United Kingdom 324; then Indonesia, Japan, Brazil and France.

Why it matters: host-nation entries at second place, and India at third, tell you generative filmmaking is spreading fastest where conventional film finance is hardest to get.

Kazakhstan sending 829 films to its own festival is partly a home-crowd effect and partly the point of the exercise: Astana Hub has spent years trying to build a technology sector, and a film festival is an unusually visible way to advertise it. India at around 700 is the more telling figure. So is Indonesia in the top ten.

The UK’s 324 is a decent showing for a country of its size, and worth holding onto the next time someone claims British film has ignored this entirely. It has not. It has just been doing it outside the institutions — there is no BFI strand for this, no Film4 development slate, no obvious route from a generative short to a commissioned project. Three hundred and twenty-four people entered anyway.

Two notes of caution on the country table. The figures published by the festival and those reported by The Astana Times differ by a few dozen films in several rows, which is normal for numbers compiled mid-process. And a submission count is not a quality count.

Photo: Ana Las Heras, via Wikimedia Commons (CC BY-SA 4.0).

6. 80+ Models, 19 Genres and One Four-Hour Film

What makes it stand out: the technical footprint of the submissions is the most useful public snapshot yet of what generative filmmakers actually use.

Key data: more than 80 generative models used across entries; 19 genres in the open competition; estimated $4 million-plus spent by entrants on tools; combined running time above 80,000 minutes by the festival’s count and above 100,000 by The Astana Times’ count; longest single submission 4 hours 3 minutes.

Why it matters: nobody has had a dataset like this before.

The four-hour entry is a rules curiosity as much as anything — the competition cap was ten minutes, so a four-hour piece sits outside the judged categories and reads more like a statement. One project reportedly spans five separate fictional universes. This is what happens when you remove the cost of a shooting day: people make things that would be insane to attempt with a crew.

The $4 million figure is the one worth sitting with. That is entrants’ own money, spent on subscriptions and credits, chasing a $1 million prize pool. As an industry economic model that is closer to a lottery than a film fund, and it is a fair criticism of the whole format. The counter-argument is that the tools were bought to make films with, not solely to enter a competition, and most of them stay useful afterwards.

7. Who Is Actually Paying For This

What makes it stand out: the backers are a technology cluster, a brokerage, a ByteDance division and a crypto-adjacent investor — not a single film body among them.

Key data: organised by Astana Hub as part of Kazakhstan’s AI Month; general partner Freedom Holding Corp; technology partner BytePlus (ByteDance); $1m production fund from Balaji Srinivasan via Network School; co-founders Almaz Zhali (board chairman) and Aizatulla Khussain (creative director).

Why it matters: it tells you what the festival is for, and it is not primarily cinema.

Kazakhstan has been pushing hard to position Astana as a technology hub between Europe and China, and AI Month is the umbrella for that campaign. A film festival is a good way to get 8,000 people in 125 countries to spend three months making content with your city’s name attached. That is not a criticism — Cannes exists partly to sell the Riviera, and Sundance transformed a ski town — but it does explain the budget.

BytePlus as technology partner is the detail that raises the most eyebrows in industry circles, given ByteDance’s parallel interests in short-form video and generative tooling. Freedom Holding as general partner puts a listed financial services group behind the money. And Srinivasan’s Network School running the production fund selections separately from the jury means there are effectively two competitions happening at once: one for the prizes, one for the investment.

Photo: Houssem Abida, via Wikimedia Commons (CC BY-SA 4.0).

8. Every Entry Gets a Plagiarism Check

What makes it stand out: mandatory screening of all 8,067 films, in a field where provenance is the central unresolved problem.

Key data: plagiarism checks compulsory for every submission; around 150 entries advance to senior review; 25 finalists; entries capped at ten minutes; generative AI required as the primary production tool.

Why it matters: it is the first large AI festival to make originality checking a formal stage rather than an afterthought.

It is also, inevitably, an incomplete answer. A plagiarism check catches a film that copies another film. It does not address the deeper objection: that the models themselves were trained on work whose creators were neither asked nor paid. No festival rule can fix that, and pretending otherwise would be dishonest. What the check does do is stop the most obvious abuse, and establish a norm that the next event will find hard to drop.

The ten-minute cap matters more than it sounds. It is set roughly at the limit of what current tools can hold together coherently, which is why almost every serious AI film competition lands in the same range. Push past it and the seams show: a face drifts, a coat changes colour, a room rearranges itself between cuts. Ten minutes is where craft still beats compute.

The funnel is brutal, too. Roughly 150 films out of 8,067 reach senior review — under two per cent — and 25 make the final. For most entrants the festival was a deadline and a reason to finish something, which is the honest function of nearly every open-submission competition ever run.

9. Why a Festival in Kazakhstan Matters to British Film

What makes it stand out: it lands in the middle of an unresolved UK argument about AI, copyright and performers’ rights.

Key data: 324 UK entries; UK government dropped its preferred copyright opt-out exception in March 2026; Academy ruled AI-generated performances and screenplays Oscar-ineligible in May 2026.

Why it matters: British filmmakers are entering a global AI competition while their own industry is still deciding what the rules are.

The UK position shifted sharply this year. Ministers had favoured an exception letting AI developers train on copyrighted material unless rights holders opted out; in March 2026 that was abandoned. Bectu head Philippa Childs called the opt-out “unworkable, unfair” and said it disproportionately harmed freelancers. Equity general secretary Paul W Fleming described the reversal as heading off “an act of national self-sabotage”. The government has signalled protections for digital replicas — unauthorised use of a performer’s voice or likeness — but has not landed on a copyright framework.

Meanwhile the Academy drew its own line in May, ruling that AI-generated performances and screenplays cannot be nominated: performances must be credited in a film’s legal billing and demonstrably performed by consenting humans, screenplays must be human-authored, and the Academy can demand documentation of a production’s AI use. Generative tools elsewhere in the pipeline remain permitted. We unpicked how that plays out for performers in AI Actors: What’s Actually Happening.

So a British entrant at Astana is working in a space where the domestic industry has no funding route, the biggest awards body has ruled part of the craft ineligible, and the legal basis of the tools is openly contested. That is an uncomfortable position, and it is worth naming rather than cheerleading past.

Photo: Gnangarra, via Wikimedia Commons (CC BY-SA 4.0).

10. How It Compares With Every Other AI Film Prize

What makes it stand out: on money and volume, nothing else is close.

Key data: Astana $2m fund and 8,067 entries; UAE Government Media Office AI Film Award $1m, 3,500 submissions, won by Zoubeir Jlassi’s Lily in January 2026; Runway’s AIFF purse quoted at $135,000-plus across six tracks in 2026.

Why it matters: the centre of gravity in AI filmmaking prizes has moved decisively away from the companies that make the tools.

Runway’s AI Film Festival has the industry credibility — Roger Avary and Jane Rosenthal have sat on its jury, and its 2026 edition expanded beyond film into new media, gaming, design, advertising and fashion, with events in New York, Los Angeles and Tokyo. But it is a tool-maker’s festival, and its prizes are partly paid in credits for its own platform.

The UAE award, run with Google Gemini and handed out at Dubai’s 1 Billion Followers Summit in January, held the record for the largest single AI film prize at $1 million, from 3,500 submissions assessed by a jury of 40 across more than 400 hours of viewing. Astana has taken both records at once.

Photo: Ken and Nyetta, via Wikimedia Commons (CC BY 2.0).

What Happens on 3 October

Twenty-five finalists, ten of them competing for a $450,000 cheque, screen in Astana across three days, and the awards go out on the final night. The jury list is still to be confirmed publicly at the time of writing; the production fund selections are made separately.

The real test is not the ceremony. It is whether any of the 8,067 films turns out to be worth watching twice — and whether the $1 million production fund produces anything that reaches an audience outside the AI community that made it. On that, the festival deserves to be judged the same way any other is: by what screens afterwards.

We will cover the winners and the finalists in AI & Future of Film as they are announced.

Sources

• Astana AI Film Festival: official site, submission statistics and prize structure
• The Astana Times: Festival attracts more than 8,000 entries from 125 countries
• The Astana Times: Festival opens submissions, offers $1 million prize fund
• DKNews: Filmmakers compete for $2 million at the Astana AI Film Festival
• Runway AI Film Festival 2026: prizes and categories
• Variety: Oscar rule changes on AI, acting nominations and international film
• C21Media: Bectu and Equity welcome UK government U-turn on AI copyright plans

Frequently Asked Questions

📸 Follow us on Instagram for daily film, TV & gaming news: @filmoriauk