Model

MiniMax's partners now sell its video model for half the lab's own price

On 2 October, HeyGen and Creatify both shipped video models fine-tuned from MiniMax's open H3 weights, at 1.5 to 4 cents a second against the lab's own 8-cent list. The price floor in AI video was set this week by resellers of open weights, not by the labs that train them.

Vincent JiangVincent Jiangtwo-minute read
Share
Joshua Xu, HeyGen's co-founder and chief executive, speaking on stage at Collision 2024
1 / 6Slide 1 of 6
HeyGen's co-founder and chief executive Joshua Xu at Collision 2024. His company shipped heygen-video-1, built on MiniMax's open H3 weights, on 2 October.

Two US launches, one Shanghai engine

On 2 October two American companies shipped video models built on someone else's weights. Creatify Labs launched Boreal-H3, an ad model "post-training MiniMax H3 in partnership with MiniMax," at $0.04 a second at 768p 1. The same day HeyGen~$500M — HeyGen, private, latest valuation ~$500M released heygen-video-1, its first general-purpose video model, which its own documentation describes as "built on MiniMax H3 and post-trained by HeyGen," turning out 5-to-15-second clips with their own audio 23. MiniMax published the complete H3 weights for fine-tuning under a community licence in August 5.

Three cents against forty-seven

At regular rates, the new H3 resellers sit at 3 to 4 cents a second at 768p. MiniMax's own API lists at 8 cents, Google's Veo 3.1 Lite at 5 cents at 720p, and Seedance 2.5 at roughly 47 cents 124. HeyGen's 1.5-cent launch rate runs through 31 October 2.

H3 resellers sell at 3 to 4 cents a second, half MiniMax's own 8

  • Estimate
0¢10¢20¢30¢40¢50¢HeyGen Video, 768p (regular)3¢Boreal-H3, 768p4¢MiniMax H3 list, 768p8¢the lab's own listVeo 3.1 Fast, 720p10¢Veo 3.1, 720p40¢Seedance 2.5 (Creatify's figure)47¢
Data
Value
HeyGen Video, 768p (regular) (estimate)3¢
Boreal-H3, 768p4¢
MiniMax H3 list, 768p8¢
Veo 3.1 Fast, 720p10¢
Veo 3.1, 720p40¢
Seedance 2.5 (Creatify's figure) (estimate)47¢
List price per second of generated video at the 720 to 768p class, US cents. HeyGen regular rate and the Seedance 2.5 figure are third-party reported; Veo rates from Google's price sheet.1,2,4

The lab's partners sell under its own list price

MiniMax Group listed in Hong Kong in January and closed 2025 with $79.0 million of revenue against a $1.87 billion net loss 6. Open weights bought the lab distribution through partners like these; the same release turns its 8-cent list price into a ceiling those partners now sell under 15. Any per-second video revenue modeled at frontier rates now has a 4-cent substitute underneath it.

The benchmark grades itself

The quality case is Creatify's own exam: 133.3 for Boreal-H3 against H3's 100 on its Ads Quality Index, scored by judges Creatify validated itself 1. The release concedes the scores measure ad quality before launch, not whether the ads convert 1. HeyGen's headline price is a sale, and its regular rate comes through a single trade report 2.

Watch 1 November

The tell is 1 November, when HeyGen's launch rate expires and its 768p price roughly doubles 2. The question for MiniMax is whether it cuts its own list or lets partners own the cheap tier. Open weights bought MiniMax an ecosystem; the invoice is a ceiling on its own API.

Get Model by email, free

The AI labs, their models, and the business behind them. In our daily email, every weekday.

Reader comments

0 comments

    Model

    HeyGen sells Avatar 6 while Tavus's Griffin scores 3.83 against a human's 3.92 and still has no price

    HeyGen put Avatar 6 on general availability on November 17 with a public API and first place in New Measure's blind Avatar Arena; Tavus's Griffin, which Tavus says is the highest score ever recorded for an AI on NVIDIA's face-to-face benchmark, is still a preview with no price. Revenue accrues to HeyGen today because Avatar 6 is the model on sale; Tavus's best model is not.

    Model

    Pika's Enact lets creators direct AI video mid-scene, with answers in under a second

    Enact, posted November 18, lets a creator drag an AI actor mid-scene, talk to the cast by name and redo the camera after locking a performance. Pika open-sourced the model underneath and kept the directing layer closed, a bet that value in AI video is moving from model quality to workflow.

    Model

    Anthropic falls 11% toward its $83 IPO price as Microsoft undercuts Claude

    Anthropic fell about 11% Friday morning, back toward the $83 IPO price, as Microsoft's price attack on Claude met the open market. Microsoft holds Anthropic's compute contract, its shares and a cheaper rival product; a leveraged income ETF on the two-day-old stock listed the same morning.

    Model

    Ramp's invoices show Anthropic held 58.7% of US business AI spend vs OpenAI's 37.2% before its record IPO

    Ramp's index of 85,000 US business bill-payers is the first independent check on the record $81.3 billion IPO: the enterprise lead over OpenAI is real, and so is the falling price of what Anthropic sells.

    Model

    Google's Gemini 4 Neon took the coding lead the day before Anthropic's $1.5 trillion debut

    Google's mid-tier model debuted at No. 1 on Artificial Analysis's coding-agent index at $6.86 a task against Claude Opus 5.5's $12.61, a day before Anthropic's Nasdaq debut. A $1.5 trillion valuation built on coding-led growth now meets a measured, cheaper rival, and the exposure sits with whoever paid $83.

    Model

    ElevenLabs, valued at $22 billion, puts its voice agents inside NICE, Genesys, Amazon Connect and Five9

    Salesforce and Five9 have put more than $3 billion into buying customer-service AI this month. ElevenLabs, valued at $22 billion after a September share sale, embedded its agents natively into Five9's platform and three rivals instead, and the evidence points to pricing power sitting with the channel, not the model maker.

    Subscribe to TheInference for free