/week free
Updated July 2026 — we add new rows as models are released, so bookmark this page.
AI model releases have settled into a relentless rhythm: every few weeks a lab ships something that changes which tool you should reach for. This page is our running tracker of the releases that actually matter — text, image, video, and speech — with a plain-language note on why each one is worth knowing about. Every model listed here that we host is live on our models page with current pricing, so you can go from reading about a model to using it in one click.
The release tracker
Major AI model releases — updated regularly
| Model | Developer | Type | Released | Why it matters |
|---|---|---|---|---|
| Claude Fable 5 / Mythos 5 | Anthropic | Text | Mid 2026 | First of the Claude 5 family; Mythos is a new top tier above Opus, with Fable as the generally available version |
| Grok Imagine 1.0 | xAI | Image + Video | 2026 | xAI's image and video generation goes public, with native image editing |
| Seedance 2.0 Fast | ByteDance | Video | 2026 | Fast, affordable text-to-video that made short clips practical at scale |
| Gemini 3.5 Flash | Text | 2026 | The workhorse fast model — strong quality at commodity pricing | |
| Seedream 5.0 Lite | ByteDance | Image | Jan 2026 | Lighter, cheaper Seedream generation with strong image-to-image |
| Claude Opus 4.8 | Anthropic | Text | Early 2026 | Anthropic's flagship before the Claude 5 era; still a top pick for hard reasoning |
| Nano Banana Pro | Image | Late 2025 | Gemini 3 Pro image model — exceptional identity preservation for photo editing | |
| GPT-5.2 | OpenAI | Text | Late 2025 | OpenAI's mainline update, strong at structured output |
| Seedream 4.5 | ByteDance | Image | Nov 2025 | Big jump in prompt adherence and multi-image reference support |
| Claude Haiku 4.5 | Anthropic | Text | Oct 2025 | Small, fast, cheap — the volume-work model |
| Nano Banana 2 | Image | 2025 | Gemini 3.1 Flash image — fast editing-capable image generation | |
| Chirp 3 HD | Speech | 2025 | Natural HD voices that raised the bar for affordable TTS | |
| Kling v2.1 | Kuaishou | Video | 2025 | Image-to-video with strong motion consistency |
| Wan / Z-Image family | Alibaba | Image + Video | 2025 | Alibaba's open-model push, competitive quality at low cost |
The current frontier: the Claude 5 family and the Mythos tier
The most-searched model story right now is Anthropic's Claude 5 launch — specifically the new Mythos-class tier. The short version: Claude Fable 5 and Claude Mythos 5 share the same underlying model, sitting above Opus in capability. Fable 5 is the generally available version with additional safeguards on dual-use capabilities; Mythos 5 is offered without those measures to approved organizations only. It's the first time Anthropic has split availability by tier this explicitly, and it signals where frontier-model access is heading: capability and availability becoming separate dials.
Image models: the identity-preservation race
The image story of the past year is less about raw beauty and more about control — keeping a specific person's face while changing everything around them. Nano Banana Pro and Seedream 4.5/5.0 lead here, which is exactly what powers tools like our AI headshot generator and outfit changer. If you're choosing an image model in 2026, the question isn't "which makes the prettiest picture" but "which respects my reference image."
How to actually try these
Reading benchmarks is one thing; generating with your own prompts is the real test. Every hosted model on this list is available through our generators — image, video, speech, and text across dozens of tools — with per-model pricing listed on the models page. Most are free to try without an account.
How this page works
We add rows when a release is real and usable — not on announcement day, but when you can actually generate with it. Dates reflect general availability. Spotted a release we're missing? Tell us and we'll add it.
