OpenCode Integrations

Bevel is a vendor-agnostic, Git-backed control plane for enterprise AI agents, where an organization’s agents, context, skills, tools, permissions, and identities are defined as files the company owns in its own infrastructure and served to any agent runtime over MCP. Context is stored as typed knowledge nodes with provenance for every fact, including where it came from, who last changed it, and when it was verified, then compiled into a graph that can be traversed, updated, and used for dashboards. Skills are written as plain Markdown procedures that process owners can read, review in diffs, and port across runtimes. Tool manifests define available capabilities, while secrets stay in a vault and access rules determine which agents may read specific files or call endpoints. Each agent has its own identity, credentials, and scope so actions remain attributable.

GPT-5.6 Sol Ultrafast is a new OpenAI API service tier that runs GPT-5.6 Sol up to 14× faster than Standard processing, bringing frontier intelligence to products and workflows where every second matters. Powered by Cerebras, it can generate up to 750 output tokens per second, allowing advanced reasoning to operate at real-time speeds without requiring a smaller or more specialized model. It is designed for time-sensitive business workflows where faster responses can change what AI can realistically do. Applications include incident response, where models can analyze logs, code changes, traces, and engineer reports while an outage is unfolding; financial research and security, where changing market signals and suspicious transactions can be assessed quickly; and customer support and voice, where complex issues can be resolved without interrupting a live conversation. In commerce, it can answer product questions, check inventory, and personalize recommendations.

Peezy Gateway is an AI inference gateway built to give developers and coding agents one endpoint for accessing frontier open models without relying on layers of third-party routing. The service is OpenAI-compatible, making it possible to point existing OpenAI SDKs, command-line agents, and other compatible tools at a single base URL instead of integrating each model provider separately. P0 is rebuilding the gateway on infrastructure it operates itself, with open models served directly from its own GPU clusters rather than through middlemen. The planned infrastructure includes B200 and B300 GPU clusters in private facilities across Singapore and China, with the goal of creating a fast, direct route to every supported model. Existing p0ag_ API keys and account credits are designed to carry over through the infrastructure migration, so current integrations do not need to start over when the gateway relaunches.

ClawTab manages Claude Code, Codex, OpenCode and shell jobs in durable tmux sessions. A headless daemon handles schedules, agent status, task titles, question detection, notifications and remote connections even when the desktop app is closed. Use the terminal tools, tmux plugin or macOS desktop app to manage agents. Open the same live sessions from iPhone, iPad or the web to follow output, answer questions and start work on your Mac or a paired Linux host. The apps and local tools are free and MIT licensed. Self-host the relay for free or use hosted Remote for $4.99 per month.

Automater is a desktop companion for developers who use several AI coding assistants and agents. It reads recognized transcript files that those tools save locally, organizes session histories into a searchable library, and displays agent activity, sessions waiting for attention, and usage indicators across supported providers in a system-tray panel. Automater Lite is available for Windows, macOS on Apple silicon, and Linux. Its local archive is stored on the user's computer, and resuming a session is available where the original tool supports it. The free Lite edition works without an account; the paid PRO edition adds managed sessions, agent controls, and fleet features. It is designed to help developers keep track of work across multiple coding tools, rather than replace those tools or provide the coding agents themselves.

Light Studio is a desktop photo-library and RAW-development application for photographers and creators who organize and edit image collections. It opens Lightroom Classic catalogs in place or converts them to Light catalogs, and supports Lightroom presets and XMP sidecars. Users can browse by date, folder and album; search photos; group faces; manage keywords and metadata; and batch-rename files. Editing tools include exposure, color, curves, grading, detail, optics and crop, plus HDR and panorama merges. Local masks, denoise, super resolution, and remove, heal and clone tools support further adjustments, while compatible video files can live alongside photos. Users can send photos to Photon Studio for retouching and return them to the library. Libraries, edits, face groups, search indexes and local AI processing stay on the computer; Light Studio is planned for macOS, Windows and Linux.

Claude for Legal is an AI assistant from Anthropic that legal professionals can use to work with documents and research materials. It can summarize lengthy files, extract and compare information, draft and revise text, and assist with contract and document analysis. Law firms and in-house legal teams can use Claude for Legal for recurring knowledge and document workflows, providing relevant context and source materials and reviewing generated responses as part of their work. Claude for Legal is available online and through desktop and mobile apps. It supports information processing and drafting; users remain responsible for checking outputs, applying professional judgment, and making legal decisions.

GPT-6 Cyber is an advanced AI cybersecurity model developed by OpenAI to help security researchers discover, validate, and remediate software vulnerabilities. The model specializes in autonomous vulnerability research, exploit development, secure code analysis, and authorized penetration testing across complex software environments. Its reasoning capabilities allow security teams to investigate large codebases, identify previously unknown vulnerabilities, reproduce security flaws, and develop candidate patches. GPT-6 Cyber achieved a 48.5% success rate on ExploitGym, representing a 51% relative improvement over GPT-5.6 Cyber. The model includes specialized alignment training and safeguards designed to respect authorized testing boundaries, prevent unauthorized actions, and accurately report research findings. GPT-6 Cyber is initially available through OpenAI's Daybreak Red program using the API and Codex for approved cybersecurity organizations and researchers.

GPT-5.4 is an advanced artificial intelligence model developed by OpenAI to support complex professional and technical work. The model combines improvements in reasoning, coding, and agent-based workflows into a single system designed for real-world productivity tasks. GPT-5.4 can generate, analyze, and edit documents, spreadsheets, presentations, and other work outputs with greater accuracy and efficiency. It also features improved tool integration, enabling the model to interact with software environments and external tools to complete multi-step workflows. With enhanced context capabilities supporting up to one million tokens, GPT-5.4 can process and reason over very large amounts of information. The model also improves factual accuracy and reduces errors compared to earlier versions. By combining strong reasoning, coding ability, and tool use, GPT-5.4 helps users complete complex tasks faster and with fewer iterations.

Verboo Code is a coding agent with unlimited tokens at a flat price. Ten open models, most with a 1 million token context window, switchable by command, and four entry points under a single subscription: an open source CLI, a desktop app for macOS, Windows and Linux, a browser extension, and an OpenAI compatible endpoint that plugs into Cursor, VS Code and JetBrains. We do not resell anyone's API: we operate our own inference layer with a proprietary router and vLLM/SGLang orchestration. That is what makes a flat price with no hidden cap possible.

Claude Fable 5.5 is not currently an announced or released Anthropic model. Anthropic's current model documentation lists Claude Fable 5.1 as the latest Fable model alongside Claude Opus 5.5 and Claude Sonnet 5.5. Fable 5.1 remains Anthropic's premium model for demanding reasoning and long-horizon agentic workloads, with a 1-million-token context window and maximum output of 128,000 tokens. It uses adaptive thinking that is always enabled and defaults to a high reasoning effort level. Fable 5.1 is priced at $10 per million input tokens and $50 per million output tokens, with cache reads priced at $0.25 per million tokens. Anthropic has released Opus 5.5 and Sonnet 5.5, but there is currently no official model documentation, pricing, release date, benchmark data, or API identifier for Claude Fable 5.5.

AI Usage Tracker is a native macOS utility for people who monitor AI coding tools and local-model activity while working. It displays supported account limits, reset times, and agent activity in a compact notch that can sit along the top, bottom, left, or right edge of the screen. Users can review session and weekly usage, receive reset notifications, and select a tracked session to return to work when an agent needs attention. The app supports Claude, Codex, Cursor, Gemini, OpenCode, Kimi, Ollama, and LM Studio; available readings vary by provider and account. For supported local models, it can show context use and generation speed, and separate work and personal profiles can be configured for some services. The installed Mac app requires macOS 15 or later and runs on Apple silicon or Intel Macs.

Bitdefender VPN for AI Agents is an installable macOS utility that routes network requests from supported AI agents through Bitdefender VPN servers. It connects Claude Desktop, Cursor, Codex, and OpenCode through Bitdefender MCP tools. Each routed prompt receives a separate encrypted connection and isolated container, which closes when the prompt ends; users can select an exit location for regional testing. The product is intended for individual developers and freelancers using agents for web research, quality assurance, and comparison tasks. It is not a device-wide VPN: traffic outside the supported agent tools is not covered, and the VPN does not conceal prompts from the AI service provider. The beta installer requires macOS 13 or later on Apple silicon; Intel Macs are not supported. The vendor says Windows support is planned.

Modelflare is a hosted AI API gateway for developers building applications, agents, and workflows that use models from multiple providers, including GPT, Claude, Gemini, Grok, and DeepSeek. It provides a shared API key and console for OpenAI-compatible chat and Responses, Anthropic Messages, and provider-specific image and video interfaces. Developers can compare models and routing groups, configure primary routes and ordered fallback groups, or use Smart API Keys to select among available routes according to a routing strategy. API keys can be separated by project and managed with controls such as quotas, expiration dates, and model or IP restrictions. The service records request-level usage, token counts, routing details, and costs, with pricing based on the selected model, group, and applicable billing unit. Documentation includes setup guides for SDKs, coding agents, plugins, and workflow tools.

GPT-6.2 Astra is a reportedly experimental artificial intelligence model from OpenAI that may represent an upcoming advancement in the company's frontier reasoning and autonomous agent technology. The model identifier has appeared in screenshots of Codex usage analytics, suggesting that a new Astra variant may be undergoing limited testing. Building on the GPT-6 Astra family, the model is expected to focus on complex reasoning, software development, and long-running AI agent workflows. Potential capabilities include advanced code generation, debugging, technical problem-solving, and automated execution of multistep engineering tasks. The model may also improve autonomous task completion, tool use, and interaction with software development environments. OpenAI has not officially announced GPT-6.2 Astra or confirmed its technical specifications, performance benchmarks, pricing, or public availability.

Kandev is an open-source development workbench for coordinating AI coding agents and software tasks. It brings task planning, agent sessions, repository activity, code changes, and review into one workspace, where developers can run agents in parallel, organize work with Kanban and workflow steps, inspect diffs, and review changes before opening or merging pull requests. Users can connect supported coding-agent CLIs or ACP agents and work with repositories through GitHub, GitLab, Azure DevOps, Jira, Linear, and Sentry integrations. Kandev runs locally or on infrastructure the user controls, with CLI, desktop, Docker, and service deployment options. Executors include local, isolated worktree, Docker, SSH, and remote environments; MCP, webhooks, and scheduled automations can coordinate tasks. It is intended for software developers and engineering teams seeking a shared workspace for agent-assisted coding and code review.

GPT-6.2 Sol is a reportedly upcoming artificial intelligence model from OpenAI that may build on the reasoning, coding, and professional productivity capabilities of GPT-6.1 Sol. References to the model identifier have appeared in reported Google Cloud configuration records, suggesting possible testing or preparation for deployment. The model is expected to support complex reasoning, software development, research, data analysis, and automated business workflows. Potential capabilities include advanced code generation, debugging, document analysis, tool calling, and autonomous execution of multistep tasks. Building on the Sol model family, GPT-6.2 Sol may emphasize a balance between computational efficiency, response quality, and reliable task completion. OpenAI has not officially announced GPT-6.2 Sol or confirmed its technical specifications, performance benchmarks, API availability, pricing, or release schedule.

Superfactory is an AI software-development orchestration and control-plane product for engineering leaders and platform teams coordinating coding agents. It brings epics, tasks, bugs, dependencies, agent runs, worktrees, pull requests, review evidence, and merge readiness into a shared workflow. Teams can plan work on a Kanban board, assign tasks to built-in agents or automated workflows, and run agents in parallel across repositories. Separate worktrees and stacked pull requests help organize dependent changes, while task-linked diffs, logs, and agent feedback support human review before merge. Built-in roles include an epic runner, planner, pull-request review agent, and bug-fixing agent. Superfactory coordinates agents and development tools teams already use rather than replacing their coding agents or IDEs; access is offered through a design-partner pilot.

OpenAI Bel is the reported codename of an unreleased artificial intelligence foundation model associated with OpenAI. Unofficial reports describe it as a large-scale pretraining project intended to support future generations of AI models. The project has been linked in speculative reporting to OpenAI's Astra model family and potential successors to GPT-6. Some sources claim that Bel contains more than 10 trillion parameters, although this figure has not been independently verified. OpenAI has not published official technical specifications, benchmark results, or documentation confirming the project's capabilities. Bel is not currently available as a publicly documented model through ChatGPT or the OpenAI API.

Hinoki Offensive One (HO1) is a cybersecurity AI model designed for offensive-security work that supports defensive testing and research. Hinoki presents it as accessible through supported developer tools, including Claude Code, OpenAI Codex, OpenCode, Visual Studio Code, and Cursor, so security professionals can use it within coding and agent workflows. The product site describes multi-step task execution and evaluations on cybersecurity benchmarks. HO1 is intended for security researchers, penetration testers, and defenders conducting authorized assessments, vulnerability research, and security testing. Its offensive capabilities make clear authorization, safeguards, and responsible-use policies important parts of deployment. Hinoki also publishes comparisons with other cybersecurity-focused and general-purpose AI models to help technical users assess model behavior for security-related tasks.