-
Notifications
You must be signed in to change notification settings - Fork 271
Core Developer Updates 2025
Completed Features:
-
API KEY [API PREFERENCE] [GET YOUR API KEY]
- Users can create their own API key on the settings page and name it accordingly.
- The domain can recognize whether the API key exists, but it does not perform request interception.
- Update the documentation and website regarding API call examples, adding examples of how to use API keys.
- Prepare a system notification to remind users to use API keys when making API calls, to be sent out after the system goes live.
-
Free API credits for developer
- We provide a feature for developers to apply for free credits, which allows them to access specific domains offered by Gaia. Note that once a free developer account is created, it will not be eligible to participate in community reward activities, and the generated throughputs will not be counted towards scoring points. If developers want to participate in community reward activities, they need to register a new wallet account.
-
- Gaia Domain for developer API request.
- After applying for free credits, the developer can only access the Gaia Domain through API and cannot earn score rewards from their free API credits usage.
https://rustcoder.gaia.domains/v1
https://llama3b.gaia.domains/v1
https://qwen72b.gaia.domains/v1
https://qwen7b.gaia.domains/v1
-
Chat UI Improvements
- Move Domain selection to the right side of the Chat UI, display selected domain on the left. Adjust button layout for a better user interaction experience.
-
Notification
- Simple message inbox
- Add a notification system to the website, allowing operators to publish various types of system notifications to all users, such as version upgrades, new feature releases, and events, etc. Support configuration of message title, content, and click-to-jump links.
-
20+ LLM Libraries supported Reorganize the LLM Library's structure to support users selecting LLMs based on model brand, parameter size, and vectorization version, and display Domain grouping for different versions of LLMs used.
Feature completed:
- Released v0.4.15
- Upgrade to
server-assistant v0.3.5
: https://github.com/GaiaNet-AI/server-assistant/releases/tag/0.3.5- Provide system information, including CPU, GPU, RAM and OS, in
server_info
( Fixes [#144](https://github.com/GaiaNet-AI/gaianet-node/issues/144) )
- Provide system information, including CPU, GPU, RAM and OS, in
- Update the value of
domain
field inconfig.json
fromus.gaianet.network
togaia.domains
- upgrade the llama.cpp plugin version to 4273
- Upgrade to
- Release llama-api-server 0.16.0
- Refactor llama-api-server and rag-api-server to adapt to breaking changes in llama.cpp
- Support new models: Phi-4, DeepSeek-V3, Qwen2-VL, Falcon3-7B-instruct, Falcon3-10B-Instruct, and Megrez-3b-intruct
- Add the
--split-mode
CLI option - Waiting for v0.4.16 release
- Under testing: support Qwen-VL
- Chat UI [link1]
- Supports user selection of Domain and Node for interaction
- Maintains compatibility with non-Domain nodes during the transition period
- Agent Store
- Supports user Domain joining
- Direct Chat Bot interaction with Domain
- Domain API interaction support
- Agent Classification
- Configurable Domain types in operations backend
- Multiple layout style support
- Backend Configuration
- Added Domain support in Agent backend
- Usage Statistics
- Added chat usage statistics and feature toggles
- Prepared for future billing implementation with real-time balance checking
- Monitor uptime and availability of node network
- Currently 150B+ in throughput processed
- https://www.gaianet.ai/network-map
- Verifiable and provable uptime metrics
- Immediate penalties for node downtime/misbehavior
- Improve reliability
- EigenLayer slashing is the security mechanism
- Partnering with MotherDAO on agent launches. Season 0 has begun
- MotherDAO and Gaia devs have built out 3 AI agents to fully run an autonomous hackathon. The grants agent paid out 43K USDC. Additionally 3 more are being built
- 2 week sprint, Treasury Manager Agent and Community Manager Agent still needed
- In collaboration with Collabland, Lit Protocol, Ai16z, Gaia
- https://github.com/collabland/AI-Agent-Starter-Kit
- domains joining, agent interacting with domains, Domain API interaction support
- Supports user Domain joining
- Direct agent app interaction with Domain
- Domain API interaction support
- Network/ProtocolAgent store: domains joining, agent interacting with domains, Domain API interaction support
- Enhance RAG functionality and support for new models
- Gaia can now support 20+ models, not all published on the Gaia website
- Why does it matter? Task-Specific Performance, Cost and Resource Efficiency, Customization and Control
- https://huggingface.co/gaianet/Falcon3-3B-Instruct-GGUF/
- https://huggingface.co/gaianet/Falcon3-1B-Instruct-GGUF
- https://huggingface.co/gaianet/Megrez-3B-Instruct-GGUF
- https://huggingface.co/gaianet/Falcon3-7B-Instruct-GGUF
- https://huggingface.co/gaianet/Falcon3-10B-Instruct-GGUF
- https://huggingface.co/gaianet/Llama-3_1-Nemotron-51B-Instruct-GGUF
- https://huggingface.co/gaianet/OuteTTS-0.2-500M-GGUF
- Data marketplace (Vana)
- No-code GUIs
Current: 140K nodes Last week, Nodes: 87,724 to 120,141 (37%)
- Gaia node released v0.4.6
- Description:
- Use Llama-3.2-3b as the default node
- Nodes can now join a domain by modifying its confi settings
- Optimize the default config.json
- Moved the entries around so that they can be roughly divided into 4 sections: Node information, Chat LLM, Embedding model, and RAG Knowledge base
- Reduced the chat model context size so that it can run on small devices
- Used correct context size for the embedding model
- Remove calls to finish_single in
compute_stream
func for llama-api-server.wasm and rag-api-server.wasm- performance optimization (memory management)
- scale/efficiency
- handling AI computation streams
- How to Use: Download the v0.4.6 version
- Description:
- CMS tool monitors the status of nodes within the online node list
- Nodes with a 404 status will be removed from the Node List in the Chat UI, and high-quality nodes will be prioritized and listed at the top for user access
- Video translation to showcase LLM capabilities [Dub your video into various languages]
- Mantle agent Description: Mantle is a layer 2 project. How to use: https://www.gaianet.ai/chat?subdomain=mantle.us.gaianet.network
- GDN White List Phase Launch
-
Description: The GDN sales event has officially commenced, enabling users to directly search for and acquire domain names or purchase packages for redemption.
-
GDN sales begun
- 607 holders
- 76% minted
-
GDN sales begun
-
How to Use:
- To participate, navigate to the event page at https://www.gaianet.ai/gaia-domain-name and search for "Gaia Domain Name". Enter the activation code to complete your purchase.
- Alternatively, access the event page at https://www.gaianet.ai/gaia-domain-name, select a package, and enter the activation code to finalize your purchase.
- Following a successful transaction, you can access the list of purchased domain names by navigating to the "Assets-Gaia Domain Name" option within the login menu.
- Additionally, you can view records of individual name orders and package orders by accessing the "Assets-Order History" option within the login menu.
- The GDN homepage provides real-time statistics on the number of minted domain names, available discounts, and owner demographics.
-
-
Enhanced Throughputs Display in Assets
- Description: The cumulative Throughputs indicator in Assets will now display a more comprehensive set of usage data for users who have not logged into the portal, providing a more detailed understanding of their activity.
-
Agent Category Launch
- Description: Support for backend expansion of category quantity and corresponding agent addition
We have rebranded to Gaia! Check out this tweet for more information!
- Initial support for web search Description: allow gaia node to search web content
- How to Use: Waiting for the new version
- Pre-Release Gaia node version 0.4.0 Description:Migrate to WasmEdge 0.1.4.0
- Add Server Assistant component
- fix the bug: The regular expression does not fully match the output of the tool-call
- Add BreezeInstructPrompt to support the Breeze-7B-Instruct-v1_0-GGUF model.
- How to Use: curl -sSfL 'https://github.com/GaiaNet-AI/gaianet-node/releases/download/0.4.0/install.sh' | bash
- Improve Cryptobot to support web search
- Description: Cryptobot is a lightning-fast AI Chatbot that responds with live interactive stock charts, financials, news, screeners, and more. When the tool can’t return results, the cryptobot will search content on the web. How to Use: https://cryptobot-on-gaia.pages.dev/ Developer docs: https://github.com/JYC0413/cryptobot-on-gaia
- Integration with Agent Zero
- Description: Agent Zero is a general purpose AI agent application. You can simply ask it to accomplish tasks on the command line. It is designed to be dynamic, organically growing, and learning as users use it. How to Use: https://docs.gaianet.ai/user-guide/apps/agent-zero
- Integration with Cursor
- Description: Cursor is an AI-powered code editor / IDE. Using LLMs to generate and review code, Cursor is an alternative to the very popular GitHub Copilot. How to Use: https://docs.gaianet.ai/user-guide/apps/cursor
- Build an agent for GaryVee Description: Build a digital agent for GaryVee. How to Use: https://gary.gaianet.network
New LLM models
- https://huggingface.co/gaianet/Llama-3-Taiwan-70B-Instruct-GGUF
- https://huggingface.co/gaianet/Llama-3-Taiwan-70B-Instruct-GGUF
- https://huggingface.co/gaianet/Breeze-7B-Instruct-v1_0-GGUF
- xLAM-7b-fc-r-GGUF: https://huggingface.co/gaianet/xLAM-7b-fc-r-GGUF
- xLAM-8*7b-r-GGUF: https://huggingface.co/gaianet/xLAM-8x7b-r-GGUF
- xLAM-8*22b-r-GGUF: https://huggingface.co/gaianet/xLAM-8x22b-r-GGUF
- xLAM-1b-fc-r-GGUF: https://huggingface.co/gaianet/xLAM-1b-fc-r-GGUF/
What we have done A brand new website has been launched!
- Optimized Video Loading Time on Homepage
- Revised Footer and Menu Text
- Removed Link to Network Map
- Modified Welcome Page Text in Chat UI
- Corrected Logo for Trending Agent
- Resolved Missing Information in Node Info after Node Binding
- Fixed Font Display Error in Node List after Node Binding
- Resolved Bug in Email Subscription Function
- Modified Response after Successful Email Subscription
- Displayed 404 Page when Clicking on Grant Program Link in Footer
- Removed Bug Bounty from Grant Program
- Revised Search Box Prompt Text in LLM Library Section
- Fixed Endpoint Display Error in API Tutorial of Chat UI
- Released Node version 0.3.2
- Solved the CORS problem
- New llama-api-server and rag-api-server
- New CLI options:
--threads
,--grammar
, and--json-schema
- Server Assistant *push the node info to the frpc (lightweight & extensible)
Description: Stockbot is a lightning fast AI Chatbot that responds with live interactive stock charts, financials, news, screeners, and more. *How to Use:[ https://docs.gaianet.ai/user-guide/apps/codegpt](https://docs.gaianet.ai/user-guide/apps/stockbot/)
- Description: A vs code extension to help developers write code
- How to Use: https://docs.gaianet.ai/user-guide/apps/codegpt
- Description: A tool to test the performance of LLM
- How to Use: https://prompt-engineering-toolkit-rho.vercel.app/
- Description: build a digital agent for Dave Krugman.
- How to Use: https://dave.gaianet.network/?aia=38f84520bde99d0f
- Finished several components of the new website
- Developed Telegram MiniApp for Gaia Agents**
- Alaa - Danielle Do you think it would be useful for LLM libraries to include recommended GPU/hardware specs for running models smoothly? Additionally, should there be a classification of modalities (e.g., text, audio, code, images, multimodal) to help users understand the model's capabilities?
- 33Audits & dOrg, smart contracts (Carlos)
- Gui - follow-up on RLY integration mobile vs web, team vs solo
- Gui - needs the new updated GUI for chatbot experience (old one is gone, new brand)
What we have done
- Released Node version 2.5
- Upgrade the llama.cpp version to b 3463
- Fix the illegal instruction error (Addressing this critical issue enhances the node's reliability and prevents unexpected crashes, providing a smoother user experience)
- Important note: there was a conflict between the llama.cpp and the phi-3 models
- Release note: https://github.com/GaiaNet-AI/gaianet-node/releases
- By incorporating the latest version of llama.cpp, the foundation of many LLMs, Gaianet benefits from enhanced performance, bug fixes, and potential new features offered by the updated library, directly improving the quality of AI interactions
- New LLM support
- Gemma-2-2b-it: https://huggingface.co/gaianet/gemma-2-2b-it-GGUF
- Mistral-Large-Instruct: https://huggingface.co/gaianet/Mistral-Large-Instruct-2407-GGUF
- Expanding the range of available LLMs empowers users with more diverse language model options
- Integration with Llamatutor and Llama Coder
- https://docs.gaianet.ai/user-guide/apps/llamatutor
- Try it online: https://llamatutor.gaianet.ai/ and https://llamacoder.gaianet.ai/
- This integration simplifies the training process for users, accelerating development cycles and fostering experimentation with different AI models
- Support creating embeddings from a CSV file.
- https://docs.gaianet.ai/creator-guide/knowledge/csv
- Allowing users to leverage their structured data, Gaianet becomes a more versatile platform, enabling the creation of highly customized AI applications tailored to specific use cases
- Update the public nodes
- https://docs.gaianet.ai/user-guide/nodes/
- Ensuring public nodes are up-to-date with the latest version guarantees users access to the most advanced features and security enhancements
What we have done
- When creating a new conversation, do not retrieve information from the Node API to speed up the creation process.
- When deleting a conversation, clear the corresponding chat history.
- When the LLM response is incomplete, users are prevented from switching nodes.
- When a user has not created a new conversation, add a "New Chat" button on the guide page to make creating a new session easier.
- introduce GUI for faster adoption, choose the base model, add domain specific app, move into a stage
- next steps: create bounty for this GUI request, take Docker example, if core (Nov timeline), work with Cookbook.dev faster, more seamless GTM.
https://console.akash.network/ The “one click deploy” is important, and it’s what made Heroku popular. Akash was very difficult to deploy via CLI, and once there was a UI it was much easier and more popular.
The “one click deploy” is important, and it’s what made Heroku popular. Switching from CLI to UI will increase the number of node runners.
- discuss with Dabl Club on Core Eng IRL meetup
- Add a new CLI option –upgrade https://github.com/GaiaNet-AI/gaianet-node/pull/82
- Allow user to customize the chat_batch-size and embedding_batch_size https://github.com/GaiaNet-AI/gaianet-node/pull/79
- Upgrade the underlying llama.cpp to 3295
- The gaianet installer will download the new version automatically
- Support Google’s gemma-2-9b and gemma-2-27b https://github.com/GaiaNet-AI/node-configs/tree/main/gemma-2-9b-it
- Optimize the integration with Llamaindex https://github.com/GaiaNet-AI/docs/pull/15
- Integration with Groqbook — Generate an entire book with GaiaNet https://docs.gaianet.ai/user-guide/apps/llamaedgebook
- Integration with Obsidian Local GPT https://docs.gaianet.ai/user-guide/apps/obsidian
- Integration with GPTPDF https://docs.gaianet.ai/creator-guide/knowledge/pdf#tool-2-gptpdf
- Integration with Translation agent https://docs.gaianet.ai/tutorial/translator-agent
API Proxy vs Data Aggregator
- Adding load balancer for Gaianet domain
- Improve classification of model store
- Abtribum meetup in India
What we have done:
Important notice: Please use the following command line to install gaianet node.
curl -sSfL 'https://github.com/GaiaNet-AI/gaianet-node/releases/latest/download/install.sh' | bash
- Add Docker support: https://github.com/GaiaNet-AI/gaianet-node/pull/66
- Sync config.json to config_pub.json at gaianet start : https://github.com/GaiaNet-AI/gaianet-node/pull/73
- Improve the CI workflow: https://github.com/GaiaNet-AI/gaianet-node/pull/72
- Release version 0.1.0 and 0.1.1:
- https://github.com/GaiaNet-AI/gaianet-- node/releases/tag/0.1.1
- https://github.com/GaiaNet-AI/gaianet-node/releases/tag/0.1.0
- Update the uninstall.sh script: https://github.com/GaiaNet-AI/gaianet-node/releases/tag/0.1.1
- Working on the load balancer in domains
- Add new types of the agent: codestral and chemistry
- When you chat with a node, permissioned access (wallet access/aa)
- Improve the chatbot UI
- Improve the node provider list
- Improve the model store
- If you have some problems, create a GitHub Issues
- Collaborate on zkML research
- Ambassador Program
What we have done:
- Integration with Dify: you can now orchestrate your LLM apps
- Integration with Anything LLM: devs can now run a docker app and configure
- Integration with LobeChat: custombizable chatbot UI with backend LLM API
- https://docs.gaianet.ai/user-guide/apps/lobechat
- automatically fetch LLMs
- Integration with Open WebUI:
- https://docs.gaianet.ai/user-guide/apps/openwebui
- a self-hosted WebUI, using any GaiaNet node as the backend LLM API
- allows you to use your own or community GaiaNet nodes in any application built on Open WebUI
Ongoing
- Log improvement: https://github.com/GaiaNet-AI/gaianet-node/pull/63
- Add Docker support: https://github.com/GaiaNet-AI/gaianet-node/pull/66
- Add optional chat_name and embedding name: https://github.com/GaiaNet-AI/gaianet-node/pull/68
- Add uninstaller script: https://github.com/GaiaNet-AI/gaianet-node/pull/69
Tools
- Generate your database vector with one click https://tools.gaianet.xyz/
- Working on the load balancer in domains
- Bind your node with your Metamask account / 0x wallet
- Agent store page https://www.gaianet.ai/agents
- Add more language versions for the repo README
- If you like our work, give us a star https://github.com/GaiaNet-AI/gaianet-node
- If you have some problems, create a GitHub Issues
- More tools to generate vector collections from documents (eg using LangChain or LlamaIndex)
- Use cases for model finetuning and knowledge base
- If we are going to build a node that understands on-chain TX data, we need examples