How to Get GLM 5.3 Flash and DeepSeek V4 Flash Free in Verdant
Verdant is giving away GLM 5.3 Flash and DeepSeek V4 Flash for free with generous usage limits. Here's how the access and pricing work.
What is free in Verdant right now?
Verdant, a multi-agent coding app, is currently offering two newly released models, GLM 5.3 Flash from ZAI and DeepSeek V4 Flash, completely free to use. Both come with usage limits measured on a 5-hour and weekly basis rather than a hard cap, and those limits are wide enough that most individual users won’t hit them during normal daily coding work.
TL;DR
- GLM 5.3 Flash is ZAI’s newest flash model, natively multimodal, and Verdant lists it at a fraction of normal credit cost, though it’s currently marked free.
- DeepSeek V4 Flash (the 0731 version) is also free in Verdant right now, positioned as a deep reasoning and agentic model rather than a visual one.
- Both models carry 5-hour and weekly usage limits instead of a hard quota, which in practice covers most everyday agentic coding sessions.
- Verdant is a desktop multi-agent coding app that plans, codes, and verifies tasks on its own, and can run several agents in parallel.
- GLM 5.3 Flash was trained to visually inspect its own output, so it can check rendered interfaces or 3D results and correct mistakes instead of assuming the code just works.
- Verdant’s free trial gives new users 7 days and 100 credits with access to flagship models like Claude Opus 5 and GPT 5.6, and paid tiers start at $5 a month.
- Because the two flash models are free, the practical strategy is to let them handle routine work and save paid credits for harder problems that need a flagship model.
One coffee. One working app.
You bring the idea. Remy manages the project.
How does GLM 5.3 Flash work inside Verdant?
GLM 5.3 Flash is a mixture-of-experts model with 320 billion total parameters but only 18 billion active per token, which is why it runs fast inside an agent loop instead of forcing users to wait between steps. It’s MIT licensed and is the first natively multimodal model in the GLM 5 series.
The multimodal part matters for coding specifically. ZAI trained the model to look at rendered interfaces, gameplay, and 3D output and judge its own work through visual feedback. Verdant describes it as a “visual work partner” that goes beyond writing code. In practice, that means when an agent builds a front end, it can inspect the actual rendered result and fix visual issues rather than just assuming the code compiled correctly and calling it done. For anyone doing UI-heavy or front-end work with an AI agent, that self-checking loop is the more useful trait than raw speed.
What does DeepSeek V4 Flash bring to the free tier?
DeepSeek V4 Flash, the 0731 version, is the second model Verdant made free at the same time. Unlike GLM 5.3 Flash’s visual focus, this one is built around deep reasoning and agentic tasks with cost-efficient inference. It’s meant for the heavier logic and multi-step planning that agent workflows require, rather than checking pixels on a screen.
Together, the two models split the work naturally. GLM 5.3 Flash covers front-end, visual, and multimodal tasks. DeepSeek V4 Flash covers reasoning-heavy and agentic tasks. Between them, a user can plausibly handle most day-to-day coding tasks without paying for a single credit, since both models were released within the last month and are being offered at no cost during this promotional window.
What is Verdant and how does the multi-agent system work?
Verdant is a desktop app built around multi-agent coding. You give it a task, and it plans the work, writes the code, and verifies the result on its own, with the option to run multiple agents in parallel on different parts of a project. It also includes skills, plugins, a plan mode for structuring bigger tasks, and the ability to deploy cloud apps directly.
The model picker inside Verdant is where the free tier shows up. GLM 5.3 Flash appears there with a free tag and is listed at a small fraction of normal credit usage (0.03x), though right now it costs nothing to use within the stated time and weekly limits. DeepSeek V4 Flash sits alongside it under the same free arrangement.
Is Verdant worth paying for beyond the free models?
Verdant’s free trial gives new users 7 days and 100 credits, and that trial period opens up access to larger flagship models as well, including Claude Opus 5, Sonnet 5, GPT 5.6, Gemini 3.1 Pro, GLM 5.2, and Kimi K3. The trial also supports bringing your own API key or account, and includes one cloud app with 10 gigs of project storage and a 500 megabyte database at no cost.
After the trial, the paid tiers break down as follows:
- Light plan ($5/month): Eco mode for lighter monthly usage, with access to models like GPT 5.6, Luna, GLM 5.2, Kimi K3, K2.7 Code, and DeepSeek V4 Pro. Includes two cloud apps and 20 gigs of storage, with the option to add credits for Claude, GPT, or Gemini when needed.
- Starter plan ($19/month): With a limited-time bonus, this tier includes 480 credits per month instead of the base 320, and opens up Claude Fable 5, Opus 5, Sonnet 5, GPT 5.6, Gemini 3.1 Pro, Kimi K3, and GLM 5.2. Eco mode usage doesn’t consume credits on this tier.
- Pro plan ($59/month): 1,500 credits after the bonus, aimed at users with heavier workloads.
- Max plan ($179/month): 4,500 credits after the bonus, intended for intensive, high-volume agentic workflows.
- ✕a coding agent
- ✕no-code
- ✕vibe coding
- ✕a faster Cursor
The one that tells the coding agents what to build.
For comparison, Claude Pro and ChatGPT Plus each cost $20 a month with tighter usage limits and access to a single model family. Even Verdant’s $5 light plan, combined with the free GLM 5.3 Flash and DeepSeek V4 Flash access, compares favorably on raw value, especially since the free flash models can absorb most routine agentic work and leave paid credits for harder problems that need a flagship model like Opus 5 or Fable 5.
What are the limitations of the free tier?
The free access to GLM 5.3 Flash and DeepSeek V4 Flash is explicitly marked as limited time, so the zero-cost pricing and the bonus credit amounts on paid tiers could change without much notice. It’s also worth being realistic about what flash models are good for: they’re fast and efficient, not flagship-tier reasoning engines. For genuinely hard problems, larger models like Fable 5, Opus 5, or the full-size GLM 5.3 will still outperform the flash variants. The sensible approach is to treat the free models as your default for everyday coding and visual checks, and reserve paid credits or flagship models for the tasks that actually need deeper reasoning.
Frequently Asked Questions
Is GLM 5.3 Flash actually free to use in Verdant?
Yes, as of now GLM 5.3 Flash is free inside Verdant, subject to a 5-hour and weekly usage limit. Verdant normally prices it at a small fraction of standard credit cost, but the free offer removes that cost during the current promotional period.
What makes DeepSeek V4 Flash different from GLM 5.3 Flash?
DeepSeek V4 Flash is built for deep reasoning and agentic tasks with cost-efficient inference, while GLM 5.3 Flash is natively multimodal and specializes in visually inspecting rendered output like interfaces or 3D scenes. They’re complementary rather than competing options.
Do I need a paid plan to use these free models?
No. Both GLM 5.3 Flash and DeepSeek V4 Flash are accessible for free within Verdant without requiring one of the paid subscription tiers, though a paid plan or the free trial opens up access to additional flagship models like Claude Opus 5 or GPT 5.6.
How does Verdant’s pricing compare to Claude Pro or ChatGPT Plus?
Verdant’s light plan starts at $5 a month, a quarter of what Claude Pro or ChatGPT Plus cost at $20 a month each, and Verdant’s free flash models can handle much of the everyday agentic coding work without touching paid credits at all.
Will GLM 5.3 Flash and DeepSeek V4 Flash stay free forever?
There’s no guarantee. Both offers are described as limited time, so the free pricing and any bonus credits on paid plans could be adjusted or removed in the future.