Discount Aggregator Aggregator
Updated September 13, 2026, 3:58 p.m. ETJSON feedllms.txt
Latest prices
Discount Aggregator Aggregator

The cheapest place to render every AI video model.
Checked every day.

Ten API aggregators resell the same video models at very different prices. We read all of them daily, normalize everything to dollars per minute of generated 720p video, and show where each model is cheapest right now — for you, and for your AI agent.

Updated September 13, 2026, 3:58 p.m. ET 18 models 10 aggregators 91 live quotes compared Feed for agents: cheapest.json 1 day of history

AI aggregators Prices

Updated September 13, 2026, 3:58 p.m. ET · cheapest verified quote per model · USD per minute
#Engine / ModelProvider480p / minute720p / minuteMax clip (s)AudioSale / discountSale ends
🥇 MiniMax H3 Max Turbo ✓ fal.ai $0.38 $0.60768p 15 Included 75% launch sale 2026-09-14
🥈 Vidu Q2 Pro Fast ✓ Atlas Cloud $0.66 10 Silent 15% vs official
🥉 Seedance 1.5 Pro Fast ✓ Atlas Cloud $0.50 $1.08 12 Check setting 10% vs official
4 Veo 3.1 — Kie Lite mode ✓ Kie.ai $1.13 8 May vary No dated sale verified
5 MiniMax H3 Max ✓ fal.ai $0.75 $1.20768p 15 Included 75% launch sale 2026-09-14
6 Seedance 2.0 Mini ✓ Atlas Cloud $0.68 $1.45 15 Included 80% off Atlas list; 68% below official
7 Wan 2.6 Flash (silent) Runware $1.50 15 Off No dated sale verified
8 Wan 2.6 Flash ✓ Atlas Cloud $2.10 15 Included 30% vs official
9 PixVerse V6 (silent) ✓ Kie.ai $2.16 15 Off ≈20% below official PixVerse pricing
10 Veo 3.1 Fast ✓ Kie.ai $2.25 8 May vary No dated sale verified
11 PixVerse V6 ✓ Kie.ai $2.88 15 Included ≈20% below official PixVerse pricing
12 Grok Imagine 1.5 Developer ✓ Atlas Cloud $1.68 $2.94 15 Check setting 65% vs official
13 Veo 3.1 Lite Atlas Cloud $3.00 8 Included No dated sale verified
14 Wan 3.0 ✓ Replicate $1.50 $3.00 30 Check setting Live billing 50% below README list; page…
15 Seedance 2.0 Fast ✓ Atlas Cloud $1.62 $3.49 15 Included 70% off Atlas list; 52% below official
16 Wan 2.7 ✓ Kie.ai $4.80 15 Check setting No dated sale verified
17 Kling 3.0 Turbo ✓ Kie.ai $5.40 15 Check setting No dated sale verified
18 Seedance 2.5 ✓ Runware $6.15 $13.82 30 Included No dated sale verified

Sorted by 720p price — click any column header to re-sort. Model names link to the provider's page for that exact model. ✓ = price and comparability re-checked by two independent verifiers. ⚑ = flagged for review (see tooltip). 480p is the rate at the selected 720p provider; blank = not offered or not verified.

AI aggregators we check (10)

AggregatorRole in this comparisonWebsite
Atlas CloudCurrent video supplier; broad discounted cataloghttps://www.atlascloud.ai/
fal.aiCurrent video supplier; dated launch offershttps://fal.ai/
ReplicateBroad model hosthttps://replicate.com/
WaveSpeedAIVideo/image API platformhttps://wavespeed.ai/
RunwareMedia API platform; many exact tiershttps://runware.ai/
Kie.aiCredit-based aggregator; provider-specific modeshttps://kie.ai/
Hugging FacePartner inference routinghttps://huggingface.co/docs/inference-providers/index
OpenRouterUnified router; few video modelshttps://openrouter.ai/
Novita AIMulti-model API platformhttps://novita.ai/
PiAPIMulti-model API platformhttps://piapi.ai/
Footnotes
  • 480p is the rate at the selected 720p provider. Blank means unavailable or unverified; details are in the full quote matrix.
  • One minute means accumulated generated footage. It does not imply a continuous 60-second clip. Vidu uses 10s clips. Kie Veo modes use 8s clips; shorter clips raise their effective per-minute cost.
  • Seedance token-based rates are estimates. Audio, reference inputs and actual output dimensions can change a quote.
  • Provider-specific Kie Lite and Atlas Grok Developer tiers are kept separate. Added options have not been quality-tested.
  • Hugging Face routes participating providers at their rates; it is not a separate cheaper video quote. OpenRouter only lists a few video models.
  • Prices are public list prices read on the check date. Sales expire automatically the morning after their published end date.

Every model, answered in one sentence

Cheapest verified rate, what a clip costs, the full range across aggregators, and every quote we hold — one card per model. The search box above filters these too.

🥇

MiniMax H3 Max Turbo

✓ verified

As of Sep 13, 2026, the cheapest verified MiniMax H3 Max Turbo rate is $0.60/min on fal.ai (768p, audio included), so a 15-second clip costs about $0.15. Across 3 aggregators it runs $0.60–$2.40 per minute. 75% launch sale · ends 2026-09-14

$0.60per minute
$0.15per 15s clip
3aggregators
$0.60–$2.40range across aggregators
75%saved vs priciest
fal.ai $0.60Atlas Cloud $2.28Runware $2.40
cheapest source ↗ · fal.ai API docs ↗ · endpoint minimax/h3-max-turbo/image-to-video · feed id h3turbo
🥈

Vidu Q2 Pro Fast

✓ verified

As of Sep 13, 2026, the cheapest verified Vidu Q2 Pro Fast rate is $0.66/min on Atlas Cloud (720p, audio silent), so a 10-second clip costs about $0.11. Across 2 aggregators it runs $0.66–$0.82 per minute. 15% vs official

$0.66per minute
$0.11per 10s clip
2aggregators
$0.66–$0.82range across aggregators
20%saved vs priciest
Atlas Cloud $0.66WaveSpeedAI $0.82
cheapest source ↗ · Atlas Cloud API docs ↗ · endpoint vidu/q2-pro-fast/image-to-video · feed id vidu
🥉

Seedance 1.5 Pro Fast

✓ verified

As of Sep 13, 2026, the cheapest verified Seedance 1.5 Pro Fast rate is $1.08/min on Atlas Cloud (720p, audio check setting), so a 12-second clip costs about $0.22. Across 2 aggregators it runs $1.08–$2.40 per minute. 10% vs official

$1.08per minute
$0.22per 12s clip
2aggregators
$1.08–$2.40range across aggregators
55%saved vs priciest
Atlas Cloud $1.08WaveSpeedAI $2.40
cheapest source ↗ · Atlas Cloud API docs ↗ · endpoint bytedance/seedance-v1.5-pro/text-to-video-fast · feed id seed15
#4

Veo 3.1 — Kie Lite mode

✓ verified

As of Sep 13, 2026, the cheapest verified Veo 3.1 — Kie Lite mode rate is $1.13/min on Kie.ai (720p, audio may vary), so a 8-second clip costs about $0.15. Only one aggregator quotes it.

$1.13per minute
$0.15per 8s clip
1aggregator
$1.13range across aggregators
—saved vs priciest
Kie.ai $1.13
cheapest source ↗ · Kie.ai API docs ↗ · endpoint veo-3-1 (model=lite) · feed id kielite
#5

MiniMax H3 Max

✓ verified

As of Sep 13, 2026, the cheapest verified MiniMax H3 Max rate is $1.20/min on fal.ai (768p, audio included), so a 15-second clip costs about $0.30. Across 3 aggregators it runs $1.20–$4.80 per minute. 75% launch sale · ends 2026-09-14

$1.20per minute
$0.30per 15s clip
3aggregators
$1.20–$4.80range across aggregators
75%saved vs priciest
fal.ai $1.20Runware $4.80OpenRouter $4.80*
cheapest source ↗ · fal.ai API docs ↗ · endpoint minimax/h3-max/image-to-video · feed id h3max
#6

Seedance 2.0 Mini

✓ verified

As of Sep 13, 2026, the cheapest verified Seedance 2.0 Mini rate is $1.45/min on Atlas Cloud (720p, audio included), so a 15-second clip costs about $0.36. Across 8 aggregators it runs $1.45–$9.28 per minute. 80% off Atlas list; 68% below official

$1.45per minute
$0.36per 15s clip
8aggregators
$1.45–$9.28range across aggregators
84%saved vs priciest
Atlas Cloud $1.45Kie.ai $2.46OpenRouter $4.54Runware $4.86Replicate $5.40WaveSpeedAI $7.20PiAPI $8.40fal.ai $9.28
cheapest source ↗ · Atlas Cloud API docs ↗ · endpoint bytedance/seedance-2.0-mini/text-to-video · feed id mini
#7

Wan 2.6 Flash (silent)

As of Sep 13, 2026, the cheapest verified Wan 2.6 Flash (silent) rate is $1.50/min on Runware (720p, audio off), so a 15-second clip costs about $0.38. Across 5 aggregators it runs $1.50–$3.00 per minute.

$1.50per minute
$0.38per 15s clip
5aggregators
$1.50–$3.00range across aggregators
50%saved vs priciest
Runware $1.50fal.ai $1.50WaveSpeedAI $1.50Atlas Cloud $2.10Replicate $3.00
cheapest source ↗ · Runware API docs ↗ · endpoint alibaba:wan@2.6-flash · feed id wansilent
#8

Wan 2.6 Flash

✓ verified

As of Sep 13, 2026, the cheapest verified Wan 2.6 Flash rate is $2.10/min on Atlas Cloud (720p, audio included), so a 15-second clip costs about $0.53. Across 5 aggregators it runs $2.10–$3.00 per minute. 30% vs official

$2.10per minute
$0.53per 15s clip
5aggregators
$2.10–$3.00range across aggregators
30%saved vs priciest
Atlas Cloud $2.10Runware $3.00fal.ai $3.00Replicate $3.00WaveSpeedAI $3.00
cheapest source ↗ · Atlas Cloud API docs ↗ · endpoint alibaba/wan-2.6/image-to-video-flash · feed id wanflash
#9

PixVerse V6 (silent)

✓ verified

As of Sep 13, 2026, the cheapest verified PixVerse V6 (silent) rate is $2.16/min on Kie.ai (720p, audio off), so a 15-second clip costs about $0.54. Across 6 aggregators it runs $2.16–$5.40 per minute. ≈20% below official PixVerse pricing

$2.16per minute
$0.54per 15s clip
6aggregators
$2.16–$5.40range across aggregators
60%saved vs priciest
Kie.ai $2.16Runware $2.70Atlas Cloud $2.70fal.ai $2.70WaveSpeedAI $2.70Replicate $5.40
cheapest source ↗ · Kie.ai API docs ↗ · endpoint pixverse-v6/text-to-video · feed id pixsilent
#10

Veo 3.1 Fast

✓ verified

As of Sep 13, 2026, the cheapest verified Veo 3.1 Fast rate is $2.25/min on Kie.ai (720p, audio may vary), so a 8-second clip costs about $0.30. Across 8 aggregators it runs $2.25–$9.00 per minute.

$2.25per minute
$0.30per 8s clip
8aggregators
$2.25–$9.00range across aggregators
75%saved vs priciest
Kie.ai $2.25PiAPI $5.40Atlas Cloud $5.76OpenRouter $6.00fal.ai $9.00Replicate $9.00WaveSpeedAI $9.00Runware $9.00
cheapest source ↗ · Kie.ai API docs ↗ · endpoint veo-3-1 (model=fast) · feed id veofast
#11

PixVerse V6

✓ verified

As of Sep 13, 2026, the cheapest verified PixVerse V6 rate is $2.88/min on Kie.ai (720p, audio included), so a 15-second clip costs about $0.72. Across 6 aggregators it runs $2.88–$7.20 per minute. ≈20% below official PixVerse pricing

$2.88per minute
$0.72per 15s clip
6aggregators
$2.88–$7.20range across aggregators
60%saved vs priciest
Kie.ai $2.88Atlas Cloud $3.60Runware $3.60fal.ai $3.60WaveSpeedAI $3.60Replicate $7.20
cheapest source ↗ · Kie.ai API docs ↗ · endpoint pixverse-v6/text-to-video · feed id pix
#12

Grok Imagine 1.5 Developer

✓ verified

As of Sep 13, 2026, the cheapest verified Grok Imagine 1.5 Developer rate is $2.94/min on Atlas Cloud (720p, audio check setting), so a 15-second clip costs about $0.73. Only one aggregator quotes it. 65% vs official

$2.94per minute
$0.73per 15s clip
1aggregator
$2.94range across aggregators
—saved vs priciest
Atlas Cloud $2.94
cheapest source ↗ · Atlas Cloud API docs ↗ · endpoint xai/grok-imagine-video-v1.5-developer/text-to-video · feed id grok
#13

Veo 3.1 Lite

As of Sep 13, 2026, the cheapest verified Veo 3.1 Lite rate is $3.00/min on Atlas Cloud (720p, audio included), so a 8-second clip costs about $0.40. Across 6 aggregators it runs $3.00–$3.00 per minute.

$3.00per minute
$0.40per 8s clip
6aggregators
$3.00–$3.00range across aggregators
—saved vs priciest
Atlas Cloud $3.00Runware $3.00fal.ai $3.00Replicate $3.00WaveSpeedAI $3.00OpenRouter $3.00
cheapest source ↗ · Atlas Cloud API docs ↗ · endpoint google/veo3.1-lite/text-to-video · feed id veolite
#14

Wan 3.0

✓ verified

As of Sep 13, 2026, the cheapest verified Wan 3.0 rate is $3.00/min on Replicate (720p, audio check setting), so a 30-second clip costs about $1.50. Across 7 aggregators it runs $3.00–$6.00 per minute. Live billing 50% below README list; page banner says 30% off this week

$3.00per minute
$1.50per 30s clip
7aggregators
$3.00–$6.00range across aggregators
50%saved vs priciest
Replicate $3.00Atlas Cloud $4.80Kie.ai $4.80OpenRouter $5.10WaveSpeedAI $5.70fal.ai $6.00Runware $6.00
cheapest source ↗ · Replicate API docs ↗ · endpoint alibaba/wan-3 · feed id wan3
#15

Seedance 2.0 Fast

✓ verified

As of Sep 13, 2026, the cheapest verified Seedance 2.0 Fast rate is $3.49/min on Atlas Cloud (720p, audio included), so a 15-second clip costs about $0.87. Across 8 aggregators it runs $3.49–$14.51 per minute. 70% off Atlas list; 52% below official

$3.49per minute
$0.87per 15s clip
8aggregators
$3.49–$14.51range across aggregators
76%saved vs priciest
Atlas Cloud $3.49OpenRouter $5.44Kie.ai $7.44Runware $7.80Replicate $9.00PiAPI $9.60WaveSpeedAI $12.00fal.ai $14.51
cheapest source ↗ · Atlas Cloud API docs ↗ · endpoint bytedance/seedance-2.0-fast/text-to-video · feed id fast
#16

Wan 2.7

✓ verified

As of Sep 13, 2026, the cheapest verified Wan 2.7 rate is $4.80/min on Kie.ai (720p, audio check setting), so a 15-second clip costs about $1.20. Across 8 aggregators it runs $4.80–$6.00 per minute.

$4.80per minute
$1.20per 15s clip
8aggregators
$4.80–$6.00range across aggregators
20%saved vs priciest
Kie.ai $4.80Atlas Cloud $6.00Novita AI $6.00fal.ai $6.00Replicate $6.00WaveSpeedAI $6.00Runware $6.00OpenRouter $6.00
cheapest source ↗ · Kie.ai API docs ↗ · endpoint wan/2-7-image-to-video · feed id wan27
#17

Kling 3.0 Turbo

✓ verified

As of Sep 13, 2026, the cheapest verified Kling 3.0 Turbo rate is $5.40/min on Kie.ai (720p, audio check setting), so a 15-second clip costs about $1.35. Across 4 aggregators it runs $5.40–$7.80 per minute.

$5.40per minute
$1.35per 15s clip
4aggregators
$5.40–$7.80range across aggregators
20%saved vs priciest
Kie.ai $5.40Atlas Cloud $5.71Runware $6.72PiAPI $7.80*
cheapest source ↗ · Kie.ai API docs ↗ · endpoint kling/v3-turbo-text-to-video · feed id kling
#18

Seedance 2.5

✓ verified

As of Sep 13, 2026, the cheapest verified Seedance 2.5 rate is $13.82/min on Runware (720p, audio included), so a 30-second clip costs about $6.91. Across 8 aggregators it runs $13.82–$28.38 per minute.

$13.82per minute
$6.91per 30s clip
8aggregators
$13.82–$28.38range across aggregators
51%saved vs priciest
Runware $13.82OpenRouter $13.87Replicate $13.87Atlas Cloud $18.05Kie.ai $18.90WaveSpeedAI $19.44PiAPI $21.00fal.ai $28.38
cheapest source ↗ · Runware API docs ↗ · endpoint bytedance:seedance@2.5 · feed id s25

Each card has its own link (#model-<id>) — paste it to an AI agent or a colleague. * = quote excluded from ranking.

Today at a glance

The same numbers as the table, drawn so the story is obvious.

The whole market on one line

Every 720p quote we hold, placed by price (log scale). Blue = the cheapest aggregator for that model; grey = everyone else. Hover a dot to identify it.

What moved

Price changes, new winners and sales that ended — newest first. Full log: changes.json.

  • ▼2026-09-13Wan 3.0 cheapest moved Atlas Cloud $4.80 → Replicate $3.00Live research read Replicate's billing block ($0.05/s); the page's stale '50% off' wording had kept it unranked
  • ▼2026-09-13Kling 3.0 Turbo cheapest moved Atlas Cloud $5.71 → Kie.ai $5.40New quote: 18 credits/s at 720p
  • ▼2026-09-13Wan 2.7 cheapest moved Atlas Cloud $6.00 → Kie.ai $4.80New quote: 16 credits/s at 720p
  • ●2026-09-13179 quotes across 10 aggregatorsTracking started: first live edition built from a 59-agent research pass; every cheapest quote checked by two independent verifiers

Video price index

Median cheapest 720p rate across all tracked models — one reading per day.

Index today
$2.21/min
median of 18 cheapest rates
Biggest spread today
6.4×
Seedance 2.0 Mini — priciest vs cheapest aggregator

Cheapest price per model — 720p, USD per minute

Bar colour = the aggregator that wins that model today.

Every quote we found — model × aggregator

Lighter = cheaper. The ringed cell is the winner in each row. “—” = model not offered or no verifiable price. * = quote excluded from ranking (conflicting notices).

Head-to-head — compare two aggregators

Pick any two. Every model both of them sell, side by side, with the winner marked.

vs
ModelABDifferenceCheaper

Same model, different bill — cheapest vs. priciest aggregator

Why an aggregator aggregator exists: the spread on one model can be 5× or more. Log scale.

Cheapest aggregatorPriciest aggregator

Price history — cheapest 720p rate per model

One snapshot per day. Sparklines appear once there are at least two days of history.

Cost calculator

Set a model, resolution, clip length and monthly volume — every aggregator is re-ranked instantly, and you see what choosing the right one is worth.

AggregatorAudio$ / minutePer clipPer monthvs cheapest

Per-minute rates are the verified list prices at the chosen resolution; per-clip = rate × seconds ÷ 60. Clips longer than an endpoint's maximum are marked. Token-billed models are estimates for 16:9 output.

Sales & deadlines

Limited-time offers we could verify on the provider's own page. Dated sales expire automatically the morning after their end date.

ModelAggregatorOffer720p $/min nowRegularDeadline
MiniMax H3 Max Turbofal.ai75% launch sale$0.60$2.40Ends 2026-09-14 (2 days left)source
MiniMax H3 Maxfal.ai75% launch sale$1.20$4.80Ends 2026-09-14 (2 days left)source
Seedance 2.0 MiniKie.aiLimited-time reduced price$2.46Ends 2026-10-07 (24 days left)source
Seedance 2.0 FastKie.aiLimited-time reduced price$7.44Ends 2026-10-07 (24 days left)source
Wan 3.0ReplicateLive billing 50% below README list; page banner says 30% off this week$3.00$6.00No published end datesource
Standing discounts — permanent cuts and “X% below the official price” (no deadline)
ModelAggregatorDiscount720p $/minReferenceDeadline
Vidu Q2 Pro FastAtlas Cloud15% vs official$0.66$0.78No published end datesource
Seedance 1.5 Pro FastAtlas Cloud10% vs official$1.08$1.20No published end datesource
Seedance 2.0 MiniAtlas Cloud80% off Atlas list; 68% below official$1.45$7.26No published end datesource
Wan 2.6 FlashAtlas Cloud30% vs official$2.10$3.00No published end datesource
Wan 2.6 Flash (silent)Atlas Cloud-30% vs official price$2.10$3.00No published end datesource
PixVerse V6 (silent)Kie.ai≈20% below official PixVerse pricing$2.16$2.70No published end datesource
MiniMax H3 Max TurboAtlas Cloud5% vs official$2.28$2.40No published end datesource
PixVerse V6Kie.ai≈20% below official PixVerse pricing$2.88$3.60No published end datesource
Grok Imagine 1.5 DeveloperAtlas Cloud65% vs official$2.94$8.40No published end datesource
Seedance 2.0 FastAtlas Cloud70% off Atlas list; 52% below official$3.49$11.62No published end datesource
Wan 3.0Atlas Cloud20% vs official$4.80$6.00No published end datesource
Wan 3.0OpenRouter15% off$5.10$6.00No published end datesource
Wan 3.0WaveSpeedAIdiscount_rate 95 (5% off) in model data$5.70$6.00No published end datesource
Kling 3.0 TurboAtlas Cloud15% vs official$5.71$6.72No published end datesource
Seedance 2.5Atlas Cloud20% off Atlas list$18.05$22.57No published end datesource
Seedance 2.5WaveSpeedAI10% current discount$19.44$21.60No published end datesource
Other video models spotted on these aggregators (not ranked yet)
ModelCheapest on720p $/minConfidence
Runway Gen-4 Turbo (Kie runway API)Kie.ai$0.72highsource
LTX-2 19B (Lightricks/LTX-2) image-to-video via WaveSpeedHugging Face Inference Providers$0.96mediumsource
Vidu Q2 ProRunware$1.03mediumsource
Seedance 1.5 Pro (standard, silent)Kie.ai$1.05highsource
LTX-2 19B distilled (Lightricks/LTX-2) image-to-video via falHugging Face Inference Providers$1.11mediumsource
LTX-Video 0.9.8 13B distilled (Lightricks/LTX-Video-0.9.8-13B-distilled) via falHugging Face Inference Providers$1.20highsource
Vidu Q2 (standard)Atlas Cloud$1.27mediumsource
Grok Imagine Video 1.5 (Preview, standard)Kie.ai$1.35highsource
Hailuo 2.3 Standard (image-to-video)Kie.ai$1.50highsource
Hailuo 02 StandardKie.ai$1.50highsource
LongCat-Video (meituan-longcat/LongCat-Video) text-to-video via falHugging Face Inference Providers$1.50highsource
Seedance 1.5 Pro (standard)fal.ai$1.56mediumsource
Seedance 1.5 Pro (silent)Runware$1.56highsource
Hailuo 2.3 Fast (MiniMax)PiAPI$1.60highsource
MiniMax Hailuo 2.3 Fast Image to VideoNovita AI$1.90highsource
Hailuo 2.3 (MiniMax)PiAPI$2.30highsource
MiniMax H3 Open Weights (WaveSpeed-hosted)WaveSpeedAI$2.40mediumsource
Hunyuan Video 1.5WaveSpeedAI$2.40highsource
LTX-2.3Runware$2.40highsource
MiniMax H3 (Hailuo 03)Kie.ai$2.40highsource
HunyuanVideo 1.5 (tencent/HunyuanVideo-1.5) text-to-video via WaveSpeedHugging Face Inference Providers$2.40highsource
Luma Dream MachinePiAPI$2.40mediumsource
MiniMax H3 FastAtlas Cloud$2.62mediumsource
MiniMax Hailuo 2.3 Standardfal.ai$2.80highsource
Hailuo 2.3Replicate$2.80mediumsource
MiniMax Hailuo 2.3 Image to VideoNovita AI$2.80highsource
Vidu Q2 Turbofal.ai$3.00highsource
MiniMax H3 (open-weight MiniMaxAI/MiniMax-H3) image-to-video via falHugging Face Inference Providers$3.00highsource
MiniMax H3 (open-weight, PiAPI-hosted Qubico/minimax-h3)PiAPI$3.00highsource
Vidu Q3 TurboAtlas Cloud$3.06highsource
Seedance 1.5 ProRunware$3.12highsource
Hailuo 2.3 StandardAtlas Cloud$3.36mediumsource
Hailuo 2.3 I2V StandardWaveSpeedAI$3.36mediumsource
Luma Ray 3.2Replicate$3.60mediumsource
LTX 2 ProWaveSpeedAI$3.60highsource
Luma Ray3.2Runware$3.60highsource
Wan 2.2 A14B (Wan-AI/Wan2.2-T2V-A14B) text-to-video via WaveSpeedHugging Face Inference Providers$3.60mediumsource
MiniMax H3 (model value "MiniMax-H3")Novita AI$4.14mediumsource
Wan 2.6 (standard)Kie.ai$4.19highsource
MiniMax H3 (open-weight, PiAPI-hosted)PiAPI$4.20highsource
Kling 3.0 StandardAtlas Cloud$4.28highsource
HunyuanVideo 1.5 (tencent/HunyuanVideo-1.5) text-to-video via falHugging Face Inference Providers$4.50highsource
Sora 2PiAPI$4.80highsource
Grok Imagine Video 1.5 (standard, not Developer tier)Replicate$4.80highsource
MiniMax H3 (tier unlabeled)Replicate$4.80highsource
HunyuanVideo (tencent/HunyuanVideo) text-to-video via falHugging Face Inference Providers$4.80mediumsource
HunyuanVideo-I2V (tencent/HunyuanVideo-I2V) via falHugging Face Inference Providers$4.80mediumsource
Wan 2.2 A14B (Wan-AI/Wan2.2-T2V-A14B) text-to-video via falHugging Face Inference Providers$4.80highsource
Wan 2.1 I2V 14B 720P (Wan-AI/Wan2.1-I2V-14B-720P) via falHugging Face Inference Providers$4.80mediumsource
Wan 2.6 (standard, text-to-video)OpenRouter$4.80highsource
Wan 2.6 (standard, T2V / I2V)PiAPI$4.80highsource
Hailuo 2.3 ProAtlas Cloud$4.90mediumsource
MiniMax Hailuo 2.3OpenRouter$4.90highsource
Kling 3.0 Standard (Kling Video v3 Standard)fal.ai$5.04highsource
Kling 3.0 Standard (silent)Runware$5.04highsource
Kling v3.0 Standard Image-to-VideoNovita AI$5.04highsource
Kling O3 (Kling 3.0 Omni)Kie.ai$5.40highsource
Veo 3.1 Fast (silent)Replicate$6.00highsource
MiniMax H3 (official MiniMax API; tier not labeled Max/Turbo)WaveSpeedAI$6.00highsource
Wan 2.6 (standard, image-to-video)OpenRouter$6.00highsource
Wan 2.7 Text-to-VideoNovita AI$6.00highsource
Wan 2.7 Reference-to-VideoNovita AI$6.00highsource
Wan 2.7 Video EditingNovita AI$6.00highsource
Wan 2.6 Image To VideoNovita AI$6.00highsource
Wan 2.5 Image to Video PreviewNovita AI$6.00highsource
Kling 3.0 Pro (Kling Video v3 Pro)fal.ai$6.72highsource
Kling 3.0 Pro (silent)Runware$6.72highsource
Kling Video O1OpenRouter$6.72highsource
Kling v3.0 Pro Image-to-VideoNovita AI$6.72highsource
Runway Gen-4.5Replicate$7.20highsource
Wan 3.0 Video PrimeKie.ai$7.56highsource
Kling V3.0 Motion Control (Standard)Novita AI$7.56highsource
Grok Imagine 1.5 (standard)Atlas Cloud$8.40highsource
Grok Imagine 1.5 (standard, xAI)fal.ai$8.40highsource
MiniMax H3 (official MiniMax API; tier not labeled)WaveSpeedAI$8.40highsource
Kling V3 Turbo ProWaveSpeedAI$8.40highsource
Wan 3.0 PrimeOpenRouter$8.40highsource
Grok Imagine Video 1.5Runware$8.40highsource
Grok Imagine Video 1.5 (standard, xAI direct)OpenRouter$8.40highsource
xAI Grok Imagine Video v1.5 (standard tier, not Developer)WaveSpeedAI$8.44highsource
Kling 3.0 Standard + AudioPiAPI$9.00highsource
Seedance 2.0 (standard)OpenRouter$9.07highsource
Veo 3.1 QualityKie.ai$9.38highsource
Seedance 2.0Runware$9.60highsource
MiniMax H3 (open-weight MiniMaxAI/MiniMax-H3) image-to-video via WaveSpeedHugging Face Inference Providers$9.60highsource
Kling 3.0 ProOpenRouter$10.08highsource
Kling 3.0 Turbo (Pro mode)PiAPI$11.40highsource
Veo 3.1 (standard)fal.ai$12.00highsource
Veo 3.1 (silent)Replicate$12.00highsource
MiniMax H3 Turbo LoRA text-to-video via WaveSpeed (larryvrh/MiniMax-H3-Turbo-Lora)Hugging Face Inference Providers$12.00highsource
Seedance 2.0 (standard / Pro tier, task_type seedance-2)PiAPI$12.00highsource
Sora 2 ProPiAPI$14.40highsource
Veo 3.1Replicate$24.00highsource
Kling v3.0 4K Image-to-VideoNovita AI$25.20highsource
Wan 2.6 (standard, with audio)OpenRouter$2.40mediumsource
Vidu Q2 (text-to-video)WaveSpeedAI$1.88highsource
Other aggregators we looked at (candidates for the list)
AggregatorNoteBest 720p $/min seenFor model
SegmindSells Seedance 2.0 / 2.0 Mini / 2.0 Fast / 2.5, MiniMax H3 Max Turbo (768p $0.04/s, 480p $0.025/s) and PixVerse V6 (720p $0.075/s). Seedance 2.0 Mini is token-billed at $1.75/1M tokens = $0.0378/s at 720p 16:9 (page-listed) = $2.27/min. Does NOT undercut $1.45/min (Mini) or $0.60/min (H3 Max Turbo = $2.40/min at 768p, $1.50/min at 480p). Its 480p Mini rate ($0.0176/s = $1.06/min) is the only sub-benchmark number but is not 720p.$2.27Seedance 2.0 Mini
EvoLinkUnified API (170+ models). Sells Seedance 2.0 Mini (720p $0.040/s T2V/I2V, $0.025/s with video input; 480p $0.019/s - a 60%-off promo through Oct 6, 2026), Kling 3.0 Turbo (720p $0.106/s), MiniMax H3 Max (480p $0.050/s, 768p $0.080/s), H3, Seedance 2.0/2.5. Best 720p = Mini $2.40/min. Does NOT undercut $1.45/min or $0.60/min; promo expiry makes the rate unstable.$2.40Seedance 2.0 Mini
Comfy Cloud (ComfyUI Partner Nodes)Credit-billed API nodes covering every target except Wan 2.6 Flash: Seedance 2.0 Mini (1.056 cr/1K tokens ~22.8 cr/s at 720p), Seedance 2.5, Kling 3.0 Turbo (23.63 cr/s 720p), Wan 2.6, Veo 3.1 Fast (16.88 cr/s) and Lite (6.33 cr/s 720p no audio), H3 Max (15.09 cr/s 480p, 24.14 cr/s 768p), PixVerse V6 (27.16 cr/s 720p). Support KB states $1 = 211 credits, so Veo 3.1 Lite = $0.030/s = $1.80/min (matches Google list price); Mini works out to ~$6.49/min and H3 Max 768p ~$6.86/min. Nothing undercuts the benchmarks; USD figures are derived, not page-listed.$1.80Veo 3.1 Lite
DeepInfraLists PixVerse V6 T2V/I2V at a flat $0.045/s (=$2.70/min; same as WaveSpeed's 720p no-audio rate, page does not split by resolution), Veo 3.1 Fast $0.15/s, Wan 2.6 $0.10/s (no Flash), Seedance 2.0 at $4.70/1M tokens (~$0.10/s at 720p via ByteDance's W*H*24/1024 token formula). No Seedance 2.0 Mini/2.5, no Kling, no MiniMax H3. Does not undercut the benchmarks.$2.70PixVerse V6
Eachlabs (each::labs)Unified API claiming upstream rates with no markup. Seedance 2.0 Mini image-to-video page lists '$0.028-$0.045 per second depending on duration and resolution' with 720p default, i.e. $2.70/min at 720p; the text-to-video sibling page instead shows 'From $1.15' for a 10s/720p default (~$0.115/s), so the unit is ambiguous. Seedance 2.0 family, Kling and MiniMax also listed; catalog/pricing tables load client-side and could not be read. Does not undercut $1.45/min.$2.70Seedance 2.0 Mini
CometAPIRelay aggregator. Model page lists doubao-seedance-2-0-mini at $0.064/s 720p (=$3.84/min) and Seedance 2.0 at $0.1216/s 720p; its video pricing guide lists Kling 3.0 Turbo 720p w/audio $0.112/s and Veo 3.1 Fast $0.08/s, while Seedance 2.5 is 'coming soon' and Veo 3.1 Lite is not on its model pages. Does not undercut $1.45/min.$3.84Seedance 2.0 Mini
AIML API (aimlapi.com)Sells Seedance 2.0 Mini (model page: '$4.55 / gen' flat plus a 480p reference of $0.065/s, no 720p rate) and Veo 3.1 Lite ('$0.104 / sec (variable)', 480p $0.065/s). Veo 3.1 Lite at $0.104/s = $6.24/min is the only per-second figure for a target model; well above the benchmarks and roughly 3x Google's list price.$6.24Veo 3.1 Lite
MuAPIPay-per-second aggregator. Seedance 2 Mini I2V page lists $0.15/s at 720p and $0.08/s at 480p (=$9.00/min); also lists Seedance 2.5 ($0.34/s 720p), Kling v3.0 Pro Motion Control ($0.16/s), Wan 2.6, Veo 3.1, PixVerse V6 (marketing says 'from $0.033/s' but provider page shows $0.295/s) and MiniMax H3 ($1.00/s) - internally inconsistent numbers. Does not undercut the benchmarks.$9.00Seedance 2.0 Mini
EmpirioLabsPay-as-you-go aggregator. Seedance 2.0 Mini: $0.070/s 480p, $0.150/s 720p (=$9.00/min), video-input 720p $0.359/s. Kling 3.0 Turbo 720p $0.18/s; PixVerse V6 720p $0.18/s no audio, $0.24/s with audio. Clearly priced but 4-6x the tracked best; no undercut.$9.00Seedance 2.0 Mini
ApiframeCredit aggregator (1 credit = $0.01). Only target SKU with a listed price is Seedance 2.5 at $0.29/s 720p (=$17.40/min); Seedance 2.0 $0.19/s 720p, Kling 3.0 std $0.16/s, Hailuo 03 (H3, not H3 Max) $0.10/s 768p. No Seedance 2.0 Mini, Kling 3.0 Turbo, H3 Max or PixVerse V6 on its pricing page. Its own comparison table cites Atlas Cloud at $0.112/s for Seedance 2.0 720p as cheapest. Far above benchmarks.$17.40Seedance 2.5
Together AILists ByteDance Seedance 2.0 at '$0.16 / video' and Seedance 2.5 at '$0.115 / video (launching soon)', with a blanket note that displayed prices are for the lowest resolution/duration settings; no per-second or 720p rate is published, so no comparable $/min can be verified. Third-party estimates (~$0.23/s 720p for 2.5) suggest it is not a price leader. No Mini, Kling Turbo, H3 Max, Veo Lite or PixVerse V6.
Pollo AI APIUnified video API that documents Seedance 2.0 Mini (docs page says 'cheaper than Fal AI'), Seedance 2.5, Kling 3.0, Wan 2.6, Hailuo and PixVerse families. Pricing is credit-based with top-ups from $80, but pollo.ai/api-platform/pricing and api.pollo.ai/pricing returned 403 to automated fetch, so no 720p $/min could be verified. Worth a manual check in a browser session.

How it works

An aggregator of aggregators.

The problem

Video models like Seedance, Kling, Veo, Wan and MiniMax are sold through many API “aggregators” (fal.ai, Replicate, Atlas Cloud, Runware, Kie.ai and others). Each one sets its own price, runs its own launch sales, and bills in its own units — per second, per clip, per credit, per million tokens. The cheapest place for one model is often the most expensive for another, and it changes week to week.

What this site does

It reads all 10 aggregators every day, converts every quote to the same unit — USD per minute of generated footage at 720p (768p for the MiniMax H3 family) — keeps audio-on and audio-off variants separate, notes each sale and its deadline, and publishes the winner per model. The table at the top is the summary; the charts show the spread; the JSON feed lets software make the choice automatically.

The daily research run

  1. 06:17 New York time, a scheduled job wakes up (GitHub Actions, free).
  2. For every model × aggregator pair it asks the provider's own machine-readable source first — Atlas Cloud's free quote endpoint, fal.ai's and OpenRouter's model APIs, WaveSpeed's catalog, Runware's model schema, the price block embedded in Replicate and Kie.ai pages.
  3. Where no such source exists it opens the page (with a headless browser when it needs JavaScript) and lets Claude read the price for exactly that model, tier, resolution and audio setting — and, if the page moved, search the provider's site — so a redesign does not break the feed.
  4. Every figure is normalized to $/min, sanity-checked against yesterday (a 5× jump is flagged, not published), and dated sales that ended are switched to the regular price.
  5. It recomputes the cheapest aggregator per model, writes a dated snapshot for the history charts, rebuilds this page and publishes it.

First edition — built from a live multi-agent research pass; the daily robot takes over from here.

Reading the numbers honestly

“Per minute” means accumulated generated footage, not one continuous 60-second clip — the Max clip column shows the longest single clip each endpoint allows. Prices for token-billed models are estimates for a 16:9 output at the stated resolution. Provider-specific tiers (Kie's “Lite mode” for Veo, Atlas's “Developer” tier for Grok) are kept as their own rows because they are not guaranteed to be the same product as the standard API. Quality is not ranked here — only price. Every quote links to its source, carries the date it was verified, and is flagged when the robot could not re-verify it.

For AI agents & local AIs

Give your assistant this page and it never has to visit ten pricing pages again.

Endpoints

GET https://bennygoldstein.github.io/discount-aggregator/data/cheapest.jsonPer model: cheapest aggregator, API base URL, endpoint id, $/min, promo deadline. Small — fetch this.
GET https://bennygoldstein.github.io/discount-aggregator/data/prices.jsonEverything: all quotes across all aggregators, raw billing text, sources, verification verdicts, deals.
GET https://bennygoldstein.github.io/discount-aggregator/data/history/index.jsonList of daily snapshots (then history/YYYY-MM-DD.json).
GET https://bennygoldstein.github.io/discount-aggregator/data/changes.jsonWhat changed, when, and why (price moves, sales that ended).
GET https://bennygoldstein.github.io/discount-aggregator/llms.txtPlain-text instructions for language models.
https://bennygoldstein.github.io/discount-aggregator/#model-<id>Deep link to one model's answer card (e.g. #model-mini) — the plain-English sentence is in the HTML, no JavaScript needed.

Everything is static JSON over HTTPS, no key, no rate limit beyond GitHub Pages' generosity, CORS-open. It refreshes once a day; cache it for an hour and you are fine.

Routing recipe — “render this with my favourite model, wherever it's cheapest today”

Python

import requests

FEED = "https://bennygoldstein.github.io/discount-aggregator/data/cheapest.json"
feed = requests.get(FEED, timeout=10).json()

best = feed["models"]["mini"]          # Seedance 2.0 Mini (ids are in the feed)
print(best["cheapest_aggregator"],       # e.g. "Atlas Cloud"
      best["usd_per_min_720p"],          # e.g. 1.45
      best["endpoint_id"])               # the model id to pass to that API

# Now call that aggregator's API:
api = best["aggregator"]["api_base_url"]
docs = best["aggregator"]["api_docs_url"]
# → build the request per `docs`, using best["endpoint_id"]

JavaScript / Node

const FEED = "https://bennygoldstein.github.io/discount-aggregator/data/cheapest.json";
const feed = await (await fetch(FEED)).json();

const wanted = "Kling 3.0 Turbo";
const [id, best] = Object.entries(feed.models)
  .find(([, m]) => m.model === wanted);

console.log(best.cheapest_aggregator, best.usd_per_min_720p,
            best.endpoint_id, best.aggregator.api_base_url);
// estimate for a 15 s clip:
const clipUsd = best.usd_per_min_720p * 15 / 60;

Shell

curl -s https://bennygoldstein.github.io/discount-aggregator/data/cheapest.json | jq '.models | to_entries | map({model:.value.model, at:.value.cheapest_aggregator, usd_per_min:.value.usd_per_min_720p}) | sort_by(.usd_per_min)'

Drop this line into your agent's instructions

Before generating a video, fetch https://bennygoldstein.github.io/discount-aggregator/data/cheapest.json and route the job to the
aggregator listed as cheapest for the requested model. Quote the price (usd_per_min_720p × seconds / 60)
and the promo deadline if any. If the model is not in the feed, say so instead of guessing.

Model ids in the feed: