Agnes AI has announced that it is opening its omnimodal model APIs to developers and creators worldwide, free of charge and indefinitely. This is not a limited-time trial — it is genuinely free. Weekly call volume has surged to 3.12 trillion tokens over the past two weeks, and two major upgrades are arriving this week.

The Three Models Opened Up for Free

Agnes AI ranks in the global top ten on model leaderboards and is one of the few omnimodal AI Labs to appear on both the Claw-Eval and Artificial Analysis rankings. Even before going free, its prices were only about half those of comparable commercial models.

Two Major Upgrades This Week

1M Ultra-Long Context (Rolling Out Gradually)

Agnes-2.0-Flash is about to natively support an ultra-long context window of 1M tokens, and the gradual rollout has already reached 50%. No changes to existing code are needed — as long as the total content of the messages array in an API request stays within the 1M token limit, it just works.

Core use cases unlocked by 1M context:

• Project-level code understanding: stuff the entire source code, config files, and docs of a medium-to-large software project in at once for a global code review
• Whole-document reading without chunking: full-length novels, large equipment manuals, complex legal contracts — drop them in and run detail Q&A across the entire book
• Ultra-long-running Agent conversations: solves the pain point of enterprise customer service bots and virtual assistants "forgetting" or drifting in character by turn 100

4K Ultra-HD Text-to-Image (Fully Unlocked)

Agnes-Image-2.1-Flash directly unlocks 4K (up to 4096x4096) ultra-HD image output. It natively supports nearly all mainstream aspect ratios: 1:1, 3:4, 4:3, 16:9, 9:16, 2:3, 3:2, 21:9.

Integration is minimal — if you were previously generating 1K images, just change "size": "1K" to "size": "4K" and leave the rest of the code untouched. And billing for 4K generation is exactly the same as for 1K — still zero.

TTS Speech Synthesis Testing Opens Soon

Around June 19, TTS (text-to-speech) capability is expected to officially begin gradual testing. The first version offers 20 high-quality voices covering different genders, age groups, and styles, with bilingual Chinese-English generation support.

That means you can write a script with the text model, break it into storyboards with the image model, generate the visuals with the video model, and finish with voiceover from TTS — an entire AI content production pipeline, free end to end.

Open-Source Community Ecosystem

Many developers on GitHub have already built tools and adapter projects around Agnes AI on their own initiative:

• AI coding: adapted for automated coding platforms including Codex, Claude Code, OpenClaw, and OpenCode
• Workflow platforms: Hermes, WorkBuddy, and others are already integrated
• ComfyUI nodes: dedicated Agnes 4K image and video workflow nodes have appeared
• MCP services: community developers have already built an Agnes multimodal Skill

Who Is It For

• Indie developers: no multi-million-dollar budget, but wanting to compete with industry giants on creativity
• E-commerce startup teams: batch image generation, e-commerce scene swaps, high-quality creative testing
• Long-form video creators: the content production gains that come with omnimodal technology
• AI application teams: build products on free text + image + video APIs

---

Original article: Developers Are Flooding In! A Top-Ten Global AI Lab Goes Unlimited and Free, Burning Through 3.12 Trillion Tokens in a Week