Daily AI Model Rankings Update — August 29, 2026
Google just shook up the video generation leaderboard: Gemini Omni 1.1 Flash has dethroned its predecessor to claim the #1 spot in text-to-video, while also crashing into the image-to-video top 10 — and a new code generation contender called hy4-preview quietly entered the rankings. If you’re building with video or code generation APIs, today’s changes are worth your attention.
What Changed Today
- Text-to-Video — New #1:
gemini-omni-1.1-flashovertookgemini-omni-flashto become the top-ranked text-to-video model (ELO: 1515) - Text-to-Video — New in Top 10:
gemini-omni-1.1-flashentered the rankings for the first time (ELO: 1515) - Image-to-Video — New in Top 10:
gemini-omni-1.1-flashentered the rankings (ELO: 1488) - Code Generation — New in Top 10:
hy4-previewentered the rankings (ELO: 1633)
Text-to-Video: Gemini Omni 1.1 Flash Is the New King
This is a significant shakeup. Gemini Omni 1.1 Flash didn’t just enter the text-to-video leaderboard — it landed at #1 with an ELO of 1515, blowing past the previous leader Veo 3.1 Audio 1080p (ELO: 1392) by a massive 123-point margin. That’s not a marginal improvement; it’s a generational jump. The model also displaced its own predecessor, gemini-omni-flash, suggesting Google has been iterating aggressively on its omni-modal architecture.
For developers, the practical question is API access and pricing. Google’s Gemini Omni models are accessible through the Gemini API ecosystem — look for the model ID gemini-omni-1.1-flash. Pricing details haven’t been fully published yet, but given the “Flash” designation, expect this to be positioned as a fast, cost-efficient option rather than a premium tier. If you’ve been using veo-3.1 at $0.40/sec for text-to-video, this could be a meaningful cost reduction with better quality. Watch Google’s API docs closely for rate limits and availability.
FREE GUIDE
Stop Writing Design Specs by Hand
Get the free visual guide: how AI tools generate GAMP 5 documentation directly from your PLC and DCS exports. Used by Life Sciences engineers who are done doing it manually.
No spam. Unsubscribe anytime.
Image-to-Video: Gemini Omni 1.1 Flash Enters Here Too
Gemini Omni 1.1 Flash also broke into the image-to-video top 10 with an ELO of 1488, which would place it significantly above the current leader Grok Imagine Video 720p (ELO: 1402). This suggests the omni-modal approach — a single model handling text, image, and video in a unified architecture — is outperforming purpose-built video models across multiple tasks.
For builders who need both text-to-video and image-to-video capabilities, this is compelling: one model, one API integration, top-tier performance in both categories. The current image-to-video leader from xAI (grok-imagine-video-720p) still has strong community vote counts (13,668+), so it’s worth watching whether Gemini Omni 1.1 Flash maintains its lead as more votes come in. But early signals are very strong.
Code Generation: hy4-preview Enters at a Stunning ELO 1633
This one flew under the radar but might be the most consequential change today. hy4-preview entered the code generation top 10 with an ELO of 1633 — which, if the score holds up with more votes, would make it the new #1 code generation model, surpassing Claude Opus 4.6 (ELO: 1561) by 72 points. That’s a staggering margin in a category where the top models have been separated by single digits.
We don’t yet have confirmed pricing or provider details for hy4-preview, and the “preview” tag signals this is likely an early-access release. The current code generation leader, Claude Opus 4.6 from Anthropic, runs at $5/$25 per MTok via claude-opus-4-6. If you’re evaluating code generation models for production use, keep an eye on hy4-preview’s vote count — early ELO scores with limited votes can shift dramatically. But if this holds, we’re looking at a potential new top coding model that nobody saw coming.
Current Leaders at a Glance
| Category | #1 Model | Provider | ELO Score |
|---|---|---|---|
| Text-to-Video | gemini-omni-1.1-flash 🆕 | 1515 | |
| Image-to-Video | grok-imagine-video-720p* | xAI | 1402 |
| Code Generation | claude-opus-4-6* | Anthropic | 1561 |
* Note: gemini-omni-1.1-flash (ELO: 1488) and hy4-preview (ELO: 1633) have entered the top 10 in image-to-video and code generation respectively with scores above the current listed leaders. Official #1 designations may shift as vote counts stabilize. We’ll update when confirmed.
So What?
Today’s changes point to two clear trends builders should act on. First, Google’s omni-modal strategy is paying off big. A single Gemini model just took #1 in text-to-video and entered the image-to-video top 10 — if you’re building any video generation feature, you should be testing gemini-omni-1.1-flash immediately, because consolidating your video pipeline into one model simplifies your stack and likely reduces costs. Second, the code generation leaderboard may be about to flip. The hy4-preview entry at ELO 1633 is eye-popping, but it’s a preview model with an unproven vote count — don’t rip out your Claude Opus integration yet, but absolutely add it to your eval suite. The actionable move today: spin up a quick benchmark of gemini-omni-1.1-flash for your video use cases, and put hy4-preview on your watchlist for code generation. The models that matter are shifting fast, and the builders who test early ship first.
