Changelog
Follow up on the latest improvements and updates.
RSS
We've launched a host of new models on Merlin.
GPT-4.1 Series
We've added three brand-new OpenAI models:
- GPT-4.1: Upgraded version of GPT-4o with better coding skills and understanding of complex instructions. Costs 15 ⚡ per non-Live response.
- GPT-4.1-mini: Similar quality to GPT-4o but faster and cheaper. Costs 3 ⚡ per non-Live response.
- GPT-4.1-nano: Replacement for GPT-4o-mini with a 100k token context window in Large Context mode. Costs 1 ⚡ per non-Live response.
Grok Models
Two new models from xAI are now available (finally):
- Grok 3: Excellent at math, reasoning, and coding problems. Costs 25 ⚡ per response.
- Grok 3 mini: Faster, cheaper version that's still great at math and coding. Costs 2 ⚡ per response.
We have plans to increase the context window on GPT-4.1 series models soon. Let's see how that goes!
We're launching
Deep Research
(beta) - our powerful research and analysis tool accessible via getmerlin.in/chat in Labs.Key Features:
- Scrapes up to 150 sources per query
- Comparable to OpenAI's deep research agent
- Optimized for product comparisons, recommendations, and idea generation
- Hybrid architecture combining Plan-and-Execute with full agentic pipeline
Access Details:
- Available in Labs (enable manually)
- Currently limited to 5 uses/month for Pro users
Upcoming
:- Extended access for Pro users (hopefully upto 20) and Free tier availability (a lower depth version)*
- Export query as a beautifully formatted PDF
- Upload documents as context
- Use Deep Research with Bots, Crafts and Projects
🚨 ⚠️ Known Issues:
- Reports may stall after 9 minutes
- If stalled: Wait and refresh after a few minutes -- your report will be generated!
- We're actively working on fixing this issue.
Important Notes for AppSumo Users:
- Each Deep Research query costs 800 Merlin queries, so budget your queries accordingly
FAQ:
- How do I access Deep Research?
- Go to getmerlin.in/chat
- Enable Deep Research from Labs section
- Turn on the Deep Research button
- Shoot your query!
- Who can use Deep Research?
- Currently available for Pro and Lifetime Deal users
- Limited to a max of 5 uses per month
- What's the query cost?
- AppSumo users: 800 Merlin queries per Deep Research use
- Pro users: From 3 to 5 uses per month, depending on subscription price
- What happens if my report stalls?
- Wait a few minutes
- Refresh the page
- Your report should appear
We're continuously improving Deep Research to provide you with comprehensive, reliable research capabilities. Stay tuned for expanded access and stability improvements!
Hey fam,
We have a small product improvement for when you use Live Search in a chat thread.
You'll find a new Sources button on the Decision Dialog as well as on the bottom of each answer. Clicking on it reveals a sidebar with all sources and brief context about those sources.
This is helpful when you:
- want to look for product reviews
- want to search for product listings on Amazon, eBay, etc. matching a search criteria
- want reading resources on a topic to research further
Let us know!
new
Bonkers
Introducing Inpainting on Bonkers!
We're launching
Inpainting
on Bonkers. This allows you to select an area on your image and ask AI to change the image in that selection area, based on a prompt you'll give it!This feature is only available on web at getmerlin.in/bonkers-magic, and non-Free users of Merlin and Bonkers.
Here's how it works:
- Go to getmerlin.in/bonkers-magic
- Upload your own image or generate an image as usual
- Preview an image and select Magic Editor on the bottom right
- Select an area on the photo using one of two selection tools at the bottom (freeform selection or rectangular selection)
- Specify the prompt for what you want to generate in that selected area.
- You should receive an edited image now!
Please use Bonkers Magic, and feel free to report feedback!
Why should you switch to Bonkers Magic?
- Has inpainting support now!
- Has better models like Ideogram V2 Turbo and Recraft V3, and better access to these models than separate paid subscriptions of these image generating models.
- Has a feed you can use to get inspiration and repurpose images from
- Has a fresh new, dedicated Mobile app called Wallflower (available on iOS and Android)!
FAQ:
1. Query costs for inpainting?
- For lifetime users like AppSumo, the query cost is 130 queries per inpainting edit.
2. Can I upload my own images to inpaint on them!
- Yes! Click on the 📎 button on getmerlin.in/bonkers-magic input box and upload the image you want to edit.
3. Can I use Magic Editor/Inpainting on Bonkers if I am a Merlin user?
- Only if you are a Pro subscriber or a user with a Lifetime Deal.
4. How to prompt for inpainting?
- Specify what you want to generate, not what you want to change.
For example:
----
❌
change this banana to a potato
✅
a potato
✅
a hand holding a potato
----
❌
make this box circular
✅
circular box
✅
an open circular box made of cardboard
Enjoy using the new features on getmerlin.in/bonkers-magic!
new
From the Team
A little survey on Scheduled Tasks in Merlin.
Hey Merlin fam!
Request you to read this Canny post here: https://merlin-ai.canny.io/feature-requests/p/survey-scheduled-tasks
Your response would be great help to us -- thanks in advance :)
VJ from Merlin
We're excited to unveil Focus Mode, a powerful enhancement to Merlin's web search capabilities that delivers specialized search experiences tailored to different content domains!
While the standard Live Search provides generalized web search capabilities, Focus Mode implements targeted search patterns and specialized processing pipelines for specific content domains:
What's New?
Focus Mode introduces three specialized search patterns optimized for different content types:
YouTube Mode
- Searches on YouTube for video content matching your search query, and takes them as context.
- Use it for finding latest reviews about any product, podcasts that discuss a certain topic or search for things a certain content creator has said. For example: "Who won this year's MKBHD Smartphone Awards?"
Academic Mode
- Uses Semantic Scholarto find relevant research and uses them as context. Also fetches inline citations. Currently optimal for finding LATEST research, but we're fixing it for better relevance matching.
- For example, ask: "Based on latest research on LLMs, find the major innovations in tokenisation."
Social Mode
- Currently optimized for Reddit content with future Twitter integration planned.
- Use it for knowing public opinion or validation about something. For example, ask "What do people think of Airpods Max in comparison to XM5s or QC Ultras?"
How It Works
Focus Mode automatically adapts its search and processing pipeline based on your content domain:
- Select your desired Focus Mode from the Live Search toggle below the Merlin chat bar
- Enter your search query
- Let Merlin handle the search process
- Receive domain-optimized results with comprehensive citations
Frequently Asked Questions
Q: How do I access Focus Mode?
A: Focus Mode is available to all signed-in Merlin users. Find the Live Search toggle below your chat bar, and click on the dropdown to find all Focus Modes.
Q: Does Focus Mode incur extra query costs?
A: Focus Mode costs as much as a Live Search query as usual.
Q: What are the current limitations?
A: Social Mode is currently limited to Reddit content, with Twitter integration coming soon. Academic Mode may have restricted access to some paywalled papers, and currently fetches recent results better than relevant results.
Q: Can I use Focus Mode with other Merlin features?
A: Yes! Focus Mode integrates seamlessly with existing Merlin capabilities including Projects and Crafts.
Q: What happens if Semantic Scholar service is unavailable?
A: Focus Mode includes fallback mechanisms. If a primary service is unavailable, it will automatically switch to alternative sources to maintain functionality.
Focus Mode is currently in
beta
, and we're actively working on expanding its capabilities. We'd love to hear your feedback and suggestions for improvement!Thank you for being part of the Merlin family! 🚀
January 3, 2025 - Merlin HQ
Merlin is proud to announce three new additions to the product on getmerlin.in/chat:
Chatbots on Merlin
You can now create chatbots on Merlin Chat website. The bot creation process has been simplified to a three-step Bot Builder which helps you build simple inside Merlin easily.
- Give a name to your bot
- Describe what it is supposed to do
- Draft a welcome message i.e. an introduction that the bot will greet you with, when you first talk with it
Once you are done creating your chatbot, you can:
- Decide to make it public on our Community Bots or keep it private
- Trigger the bot in chat with @bot-name
- Use them on the Merlin extension too!
DeepSeek V3 model on Merlin
We've added the new open-source model by DeepSeek which:
- Matches GPT-4o's performance at a tiny fraction of the cost
- Smartly uses only 37B parameters at a time from its 671B total, making it fast and efficient
- Is the best performing open-source AI model currently available
- Excels at coding tasks and complex math
- Strong at reasoning, writing, and analysis across multiple languages
Costs just 2 queries per prompt, without Live Search and Large Context inputs.
Pro Finder
Turning on Pro Finder on a chat in Projects leads to better knowledge searching across your project if it has a large number and/or size of files.
It uses an agentic mechanism to scrape, check relevance and confirm the validity of the responses it gives.
From a query cost perspective, it may cost from 2x to 20x the cost of a
claude-3.5-sonnet
response, therefore, for non-Pro users (like AppSumo LTD users), we have kept the cost of one Pro Finder query to be 250 queries.The feature is still experimental and the cost is bound to go down as we optimise its performance.
Thanks -- that's all for this update!
December 20, 2024
- Merlin HQMerlin is proud to launch its first version of
Analyst Mode
, an amazing new way to turn sheets into actionable insights on Merlin.Analyst Mode is automatically triggered when you upload a CSV file to Merlin Chat. You can see this happening when you see a blue-colored circular badge on the right side of the Merlin Chat input box.
When you upload a CSV file and ask Merlin questions like, for example:
- "What is the average view time of users in each subscription price?"
- "Fetch all emails who have an education domain"
- "Analyse the data giving distribution across gender"
- or even "Analyse" (although you'll get better results with more specific prompting)
Merlin will write code to collect insights for you and present them in an inline manner with a visualisation and qualitative description.
Check out this shared chat I made, as a demonstrative example: https://www.getmerlin.in/chat/share/f92b9045-56f0-4903-9def-61c17562f6f1
For best results, use
GPT 4o
or Claude 3.5 Sonnet
.Start your thread with a CSV upload and Merlin will maintain context across your thread, until you upload another CSV or a large context file in the
same
thread.This allows you to run multiple models on the same thread, making the most out of each model's strengths. For example, you could:
- Upload a sales data CSV and fetch correlation between calculated metrics
- Use o1 to analyse these correlations and devise actionables, or Live Search to compare benchmark metrics from world standards
- Add these data points to a Project and chat with them repeatedly.
- Say, you want to build a dashboard for your personal projects that showcases those metrics in a dynamic way, use Crafts to build a dynamic website using React and Shadcn.
This is unparalleled and only possible in a platform that offers you all models under one hood, and we are thrilled to see what y'all can do with Merlin.
Questions?
Q. What is the size limit of CSV that can be uploaded?
-> Same as that of document uploads till date on Merlin, according to your plan.
Q. Is Analyst Mode available to everyone on Merlin?
-> Yes. Every user who is logged in.
Q. Does Analyst Mode incur extra query consumption on Free and AppSumo plans?
-> Yes. An extra 15 queries is deducted for code writing, execution and CSV analysis.
Q. Is there a usage limit for Pro users?
-> No. The only limit is the fair use policy of Merlin, which anyway exists before this.
Q. Are my CSVs safely stored?
-> Yes. We're certified on industry standards in terms of data security (ISO, SOC2, GDPR). We do not train Merlin on your CSVs, or for that matter, any of your files.
Q. What happens if I upload another CSV in the same thread if Analyst Mode was triggered before with another CSV earlier in the thread?
-> Analyst Mode will run on the context of your newly uploaded CSV and any context available as text in the conversation. So, imperfect context i.e. the previous CSV will not be taken in full.
Q. Can we run Analyst Mode on Excel or Numbers files (.xls, .xlsx, .numbers)?
-> Not as of now. If you want to use Excel or Numbers files, convert them into CSV files using a free online tool, for example, this. (This website is not affiliated with Merlin)
Q. How do I turn off the auto-trigger of Analyst Mode?
-> Go to Labs on the sidebar, find Analyst Mode and turn the switch off.
Q. I have kept Analyst Mode on because I want to use it at some places. But then it sometimes malfunctions and I lose queries. What do I do?
-> Please reach out at feedback.getmerlin.in / support@getmerlin.in only if your queries are deducted but Analyst Mode errored out in giving an output (i.e. showed something like "Network error - reach out to support").
Q. Does it work with Merlin Magic and Live Search?
-> Analyst Mode works with Merlin Magic, but any Live Search queries have to be done separately on the thread.
Analyst Mode is an experimental feature and will act unexpectedly sometimes. That's why if you feel query usage is important for you, we suggest you turn it off.
Thanks for being a part of Merlin's family! We'd love to hear from you on our feedback page, feedback.getmerlin.in!
Small update: OpenAI's new o1 model is now available on Merlin.
o1 provides much faster chain-of-thought reasoning than o1-preview, so you would not need to spend time staring at "o1 is taking time to think..."!
It is a more refined, fine-tuned version of the o1-preview model, which was also widely used by Merlin Pro users.
The baseline query cost is 180, same as that of o1-preview.
Visit getmerlin.in/chat to use the new OpenAI o1.
Watch demo (0:30): https://www.tella.tv/video/chat-sharing-1-gzm2
We've introduced Chat Sharing on Merlin Chat. Now you can:
- Share your chats rich with Crafts, Code, Images and more
- Turn on anonymity while sharing in public
- Manage your shared links at one place
Chats in Projects are not yet shareable.
Chats do not auto-update on their own. We made this choice to respect user privacy. One can manually update their shared chat by clicking on Share -> Update shared chat.
Shipped with love.
Load More
→