Skip to content

Updates

Follow OctaFuse product progress, release announcements, and implementation notes. Every post is archived on this site with its original publication date.

VERSION 2.20.0

OctaFuse 2.20.0: sign in to MCP with your own accounts, and tool calls in Request logs

The Gateway now has a built-in OAuth authorization server, so users of your MCP services sign in with your business accounts instead of pasting a key. Request logs absorb tool invocations and gain multi-select filters by type, model and provider, and search and fetch tools add a DeepSeek engine. Apply migration 0036 before upgrading.

Read the post
VERSION 2.19.0

OctaFuse 2.19.0: A redesigned Admin console and a refreshed model catalog

The Admin console gets one visual style and tidier filters, details and forms, and maintenance mode moves into Config. The catalog adds Claude Fable 5.2, GLM-5V-Turbo, Hunyuan HY4 and more. No database migration; spot-check routes and logs after upgrading.

Read the post
VERSION 2.18.0

OctaFuse 2.18.0: OAuth for MCP and visual management of sticky bindings

When you offer gateway MCP to another product, people can sign in with that product's account instead of pasting an API key. Sticky routing can show who is bound, change the lifetime, and unbind a user. Upgrade requires migrations 0034 and 0035.

Read the post
VERSION 2.17.0

OctaFuse 2.17.0: maintenance mode, request-log summaries, and split DashScope speech routes

OctaFuse Gateway 2.17.0 can reject new inference during a maintenance window and summarize request logs by provider, model, and user. DashScope speech routes are split by series. Upgrade requires migrations 0032 and 0033.

Read the post
VERSION 2.16.0

OctaFuse 2.16.0: ElevenLabs, native vendor paths, and list-based Admin

OctaFuse Gateway 2.16.0 adds ElevenLabs speech and transcription, moves DashScope, MiniMax, and Volcengine native calls onto vendor-prefixed paths, and switches provider, model, and route Admin to lists. Upgrade requires migration 0031.

Read the post
VERSION 2.15.0

OctaFuse 2.15.0: aggregated MCP, embeddings, and a manual official-catalog check

OctaFuse Gateway 2.15.0 exposes remote MCP servers through one gateway address, routes OpenAI and Gemini embeddings, and lets you compare imported models and providers with the official catalog before applying changes. Upgrade requires migration 0030.

Read the post
VERSION 2.14.0

OctaFuse 2.14.0: more image and speech routes, discounts by route group, and safer Admin saves

OctaFuse Gateway 2.14.0 routes OpenAI image and speech clients to MiniMax, Seedream, and Gemini, sets a user discount per route group, and saves only the Admin fields you changed.

Read the post
VERSION 2.13.0

OctaFuse 2.13.0: provider discounts, route tests before save, and clearer usage

OctaFuse Gateway 2.13.0 schedules provider discounts, tests an unsaved route in the editor, and splits usage into cost, performance, and reliability views.

Read the post
VERSION 2.12.0

OctaFuse 2.12.0: provider accounts, a route workspace, and multimodal model discovery

OctaFuse Gateway 2.12.0 names provider accounts, checks upstream quota in place, finds routes faster, and tells clients which image or audio endpoint a model accepts. Admin also works on a phone.

Read the post
VERSION 2.11.0

OctaFuse 2.11.0: stream guards, user discount modes, and stable error codes

OctaFuse Gateway 2.11.0 protects long text streams, can fail over when the first event never arrives, combines user discounts in two ways, and marks proxy failures with a stable error code.

Read the post
VERSION 2.10.0

OctaFuse 2.10.0: lock route parameters, show model endpoints, and audit wallet changes

OctaFuse Gateway 2.10.0 can force a route's headers or body over the client, lists the endpoints a model actually accepts, and records permanent-balance changes with richer audit filters.

Read the post
VERSION 2.9.0

OctaFuse 2.9.0: user and key rate limits, custom upstream headers, and a clearer Admin

OctaFuse Gateway 2.9.0 adds separate rate limits for a user and each API key, lets a route attach custom upstream headers, and refreshes the dashboard, usage views, and Cloudflare multi-domain setup.

Read the post
VERSION 2.8.0

OctaFuse 2.8.0: dual quota balances, unified protocol adapters, and peak/off-peak pricing

OctaFuse Gateway 2.8.0 lets subscription allowances and top-up balances work together, simplifies client-to-upstream protocol routing, and improves Alibaba Cloud Model Studio image models, Node realtime audio, and time-based pricing.

Read the post
VERSION 2.7.0

OctaFuse 2.7.0: weekday-aware pricing, per-model user factors, and DashScope multimodal ASR

OctaFuse Gateway 2.7.0 adds weekday-aware pricing schedules for DeepSeek's new weekend off-peak policy, per-model user charge factors, synchronous DashScope multimodal ASR, and clearer user and request operations.

Read the post
VERSION 2.6.0

OctaFuse 2.6.0: safer Vertex AI auth, image billing, and daily operations

OctaFuse Gateway 2.6.0 adds project-scoped Vertex AI service-account auth, explicit image billing and no-charge failure rules, daily schedule overrides, stronger Playground and Simulator workflows, and new Provider presets.

Read the post
VERSION 2.5.0

OctaFuse 2.5.0: bring the OpenAI Responses API into unified routing

OctaFuse Gateway 2.5.0 adds an OpenAI Responses-compatible endpoint, rebuilds the Routes and Provider workspaces, and expands the Gemini, GLM, Grok, and DeepSeek model catalog.

Read the post
VERSION 2.4.0

OctaFuse 2.4.0: route DashScope speech through one gateway

OctaFuse Gateway 2.4.0 adds native DashScope ASR, TTS, and realtime audio routing, easier Model Studio and Qwen Token Plan onboarding, plus named Admin API Keys.

Read the post
VERSION 2.3.0

OctaFuse 2.3.0: keep upstream selection stable across requests

OctaFuse Gateway 2.3.0 adds Route Pool Provider Sticky, binding observability and operations, plus the hash_affinity and weight_priority strategy IDs.

Read the post
VERSION 2.2.0

OctaFuse 2.2.0: routing policy for every priority tier

OctaFuse Gateway 2.2.0 unifies Gemini Generate Content routing, adds per-priority-tier Route Pool strategies, and improves Admin configuration.

Read the post
VERSION 2.1.2

OctaFuse 2.1.2: stable routing views and richer diagnostics

OctaFuse Gateway 2.1.2 improves Admin routing and Provider UX, adds external_system to request logs, and fixes clean-checkout Admin development.

Read the post
VERSION 2.1.1

OctaFuse 2.1.1: identifiable errors and more precise circuits

OctaFuse Gateway 2.1.1 unifies user+model circuits, improves Failover and error-code contracts, and adds better upstream diagnostics.

Read the post
VERSION 2.1.0

OctaFuse 2.1: bring Agent Tools under one gateway

OctaFuse Gateway 2.1 brings Web Search, Web Fetch, Deep Search, and AI Detection into one authentication, budget, pricing, logging, and audit system.

Read the post
VERSION 2.0.0

OctaFuse 2.0: make complex routing visible

OctaFuse Gateway 2.0 introduced Surface → Pool → Target topology, priority tiers, visual strategy controls, and single-key Providers.

Read the post
DEPLOYMENT

Run OctaFuse on Cloudflare without managing a server

Deploy OctaFuse with Cloudflare Workers and D1 to centralize AI Providers, API keys, budgets, billing, and request history.

Read the post
INTRODUCTION

Why we built OctaFuse: consolidate scattered token resources

OctaFuse is a self-hostable AI capability gateway and operations control plane for Providers, user keys, budgets, billing, routing, and audit.

Read the post