akash

Prop 298: Community Support for Akash Chat, Akash ChatAPI, and AkashGen

Motivation

The Akash ecosystem continues to evolve rapidly, and a key part of that growth is the deployment and scaling of community-first tools that drive real-world utility, visibility, and long-term engagement.

AkashChat, AkashChat API, and AkashGen are powerful tools that serve as on-ramps for both developers and users who want to leverage decentralized infrastructure for AI and generative services. These tools are already being used frequently, with positive feedback and measurable traction in user activity. Github Discussion.

Since the inception of AkashChat in late 2023, Overclock Labs has been supporting the development and infrastructure that powers these applications. To realize their full potential, meet the growing demand, and ensure the sustainability of the compute that powers these tools, the Overclock Labs Core Team proposes placing the responsibility for funding them on the Akash Community Pool. Support from the Akash Community Pool for this work ensures that the Akash ecosystem continues to evolve with high-demand, developer- and user-facing tools that lower the barrier to entry for decentralized computing, particularly in AI and generative services.

By supporting this proposal, the Akash community will enable a diverse range of users, from individuals, to startups and enterprise developers, to interact with and build on Akash's decentralized cloud.

AkashChat, AkashChat API, and AkashGen not only demonstrate what is possible on the Akash Network, but also provide a smooth onboarding path for both technical and non-technical users to explore, create, and develop using Akash Network's Supercloud.

Ask: $602,650.59 for a 6-month period from January, 2025 - June, 2025

To ensure this growth is facilitated in the most responsible and appropriate manner, Overclock Labs is requesting $602,650.59 from the Akash Community Spend Pool for H1 2025.

Detailed breakdowns of the budget for this proposal can be found here:

Budget

Overclock Labs is requesting funding from the Akash Community Pool to cover deployment and ongoing maintenance costs for this family of tools over an initial year, splitting the funding rounds for this effort into two halves, Half One (H1) being from January, 2025, to June, 2025.

The table below outlines the funding expended in H1 by both the cost of compute and the operational and labor costs needing to be recuperated by Overclock Labs for this effort. This table also signals for H2 2025, which Overclock Labs will request at the end of the calendar year.

Final numbers and all costs will be reported to the Akash Community. A breakdown of monthly costs is available here.

Total Community Spend Pool Costs:

H1 Costs (Actual)

Month Compute Costs (Actual) Operation & Development Costs Subtotal Total (inc. 21% tax + 8% vol buffer)
Jan 25 $131,020 $19,971.00 $150,991.00 $194,778.39
Feb 25 $47,390 $19,971.00 $67,361.00 $86,895.69
Mar 25 $49,855 $19,971.00 $69,826.00 $90,075.54
Apr 25 $66,710 $19,971.00 $86,681.00 $111,818.49
May 25 $20,020 $19,971.00 $39,991.00 $51,588.39
Jun 25 $32,350 $19,971.00 $52,321.00 $67,494.09
H1 Total $347,345 $119,826.00 $467,171 $602,650.59

H2 Costs (Projected)

Month Compute Costs Operation & Development Costs Subtotal Total
Jul 25 $57,890.84 $15,900.00 $73,790.83 $98,933.10
Aug 25 $57,890.84 $15,900.00 $73,790.83 $98,933.10
Sep 25 $57,890.84 $15,900.00 $73,790.83 $98,933.10
Oct 25 $57,890.84 $15,900.00 $73,790.83 $98,933.10
Nov 25 $57,890.84 $15,900.00 $73,790.83 $98,933.10
Dec 25 $57,890.84 $15,900.00 $73,790.83 $98,933.10
H2 Total $347,345 $95,400.00 $442,745.00 $593,598.60

Summary

Compute Costs Operation & Development Costs Subtotal Grand Total
Grand Total $694,690 $215,226 $909,916 $1,196,249

Features

Akashchat offers AI-powered conversational capabilities on decentralized compute.

Akashchat API provides a simple interface to access the leading open-source AI models, powered by the Akash Supercloud.

Akashgen extends Akash's utility into generative image AI, tapping into a high-demand creative use case while showcasing GPU capabilities on the Akash Network.

Impact

As of June 2025, usage data from both the AkashChatAPI and AkashChat interfaces highlight a clear demand for the Akash suite of generative tools, with the most high-performing models processing upwards of 4 million requests. Moreover, by the end of June, more than 10 million requests had been made using the various generative models employed by AkashChatAPI, and nearly 500,000 requests had been processed and answered by the models utilized by the AkashChat application.

These statistics not only demonstrate the immediate adoption and utilization of these applications but also underscore the necessity to maintain and continue developing their functionality.

Updates in the Last 6 Months:

Updates to AkashChat include the adoption of new cutting edge generative models including:

  • Qwen3 235B A22B
  • Qwen3 235B A22B Instruct 2507
  • Kimi-K2-Instruct
  • DeepSeek-R1-0528
  • DeepSeek-R1
  • DeepSeek-V3

Additionally, AkashChat has undergone several keystone upgrades.

  • A Self-Hosted Standalone Version has been introduced with optional Restricted Access and AkashChat API key compatibility.
  • Gateway/Load Balancer for managing LLM instances on Akash, which is directly usable with Akash Chat frontend.
  • Voice Transcription Implementation
  • Local Document Content Extraction
  • AkashGen Integration
  • Session-Based Access
  • Model-Specific Pages
  • SEO Optimization
  • Text Generation Animation for a better user experience
  • Persistent User Model Configuration
  • Complete Router Migration to rewrite from Pages Router to App Router.

The changes to AkashChat API include:

  • Model Management & Evaluation, where a user may add selected models with continuous evaluation
  • Documentation Redesign
  • Branding & SEO updates to promote brand refresh and search engine optimization
  • Tiered User Accounts for permissionless base access and extended tiers
  • Advanced Bot Prevention including implementation of enhanced security measures
  • Key Management Interface for user-facing page for API key management.

There are more upgrades to be performed to further improve the quality of life with these applications.

What's next:

  1. Enhanced User Accounts: Introducing access to premium features and increased limits for logged in users, including:
  • Advanced OCR capabilities with support for both document and image inputs
  • Potential integration with a web research agent to facilitate more efficient information gathering
  • Seamless integration with AkashML to leverage additional inference options
  1. Additions:
  • Video Generation: The ability to generate videos, potentially using AI-driven tools to simplify content creation
  • Coding Assistance: Integrating coding capabilities or assistance, which could range from code completion to generating code snippets based on user inputs
  • 3D Model Generation: Allowing users to create 3D models, which could be useful for a variety of applications including gaming, education, and product design
  • Akash Voice Assistant: Developing a voice assistant that can perform tasks, provide information and leverage the power of the Akash AI suite

If you are interested in contributing to strategy or development, please feel free to introduce yourself in the special interest group for Clients (#sig-clients).

Limited Market Impact & Transparent Reporting

Limited Market Impact

Overclock Labs will custody the requested funds in a new, distinct wallet so that funds from any other source are not commingled.

All funds will be liquidated and managed in a manner that ensures minimal impact on the market. These funds will be managed with the same care and attention as all previous Community Funding Proposals, with liquidations done in a fashion that will not adversely affect the market. In practice, the effort of this liquidation will add depth to the AKT market for buyers looking to enter.