Best Artificial Intelligence Software for Python - Page 16
View:
Compare the Top Artificial Intelligence Software that integrates with Python as of November 2026 - Page 16
Sort By:
This a list of Artificial Intelligence software that integrates with Python. Use the filters on the left to add additional filters for products that have integrations with Python. View the products that work with Python in the table below.
-
1
ESMFold2
Biohub
ESMFold2 is the successor to ESMFold, setting a new state of the art for single-sequence structure prediction and enabling the generation of new functional proteins through searching the ESMC model’s latent space. The model predicts high-resolution, all-atom 3D structures of biomolecular complexes directly from sequence, with optional multiple sequence alignment input for enhanced accuracy on challenging targets. It is designed for structure prediction using sequence and structure modalities, with ESM representations powering a series of looped folding layers and a diffusion model projecting pairwise representations to atomic-resolution predictions. ESMFold2 predicts protein structures directly from amino acid sequences and outputs comprehensive structural information, including all-atom coordinates for backbone and side chains, confidence metrics, and optional distogram predictions for detailed structural analysis.Starting Price: Free -
2
OrcaRouter
OrcaRouter
OrcaRouter is an OpenAI-compatible AI model router that sends each prompt to the right model across OpenAI, Anthropic, Gemini, DeepSeek, Qwen, Kimi, and 200+ frontier and open source models. It is built to preserve frontier answer quality while reducing AI inference spend by grading every prompt and routing hard reasoning to frontier models and routine work to lower-cost open source models. The routing is quality-graded, never a blind, cheap-model swap, and each request shows the difficulty grade, selected model, provider, and cost so routes are visible, auditable, and reproducible. Developers can switch by changing the API base URL, while existing SDKs, model names, and streaming behavior continue to work as before. OrcaRouter supports automatic failover, so if a provider goes down mid-stream, traffic can switch transparently, and the application avoids user-facing errors. It also includes API key management with spend caps, model allowlists, rate limits, budget enforcement, and more.Starting Price: $29 per month -
3
AG2
AG2
AG2 is the open source AgentOS for building production-ready AI agents and multi-agent systems in minutes, not months. Formerly AutoGen, it provides an open source Python framework for building, orchestrating, and scaling AI agents that can collaborate through shared context, use tools, execute workflows, and support both autonomous and human-in-the-loop patterns. AG2 is designed for developers who want to build systems, not prompts, with simple and intuitive syntax, built-in conversation patterns, and a flexible platform for multi-agent automation. Agents in AG2 can extend their capabilities with tools, allowing them to interact with external systems, fetch real-time data, execute code, search the web, process documents, and complete complex tasks beyond a model’s internal knowledge. It supports many LLM providers and local models, including OpenAI-compatible endpoints, Anthropic Claude, Gemini through Vertex AI, DeepSeek, and LM Studio.Starting Price: Free -
4
Agent Client Protocol (ACP)
Agent Client Protocol (ACP)
The Agent Client Protocol (ACP) standardizes communication between code editors, IDEs, and coding agents, making agent-editor interoperability the default instead of requiring custom integrations for every possible combination. It provides a standard interface for communication between AI agents and client applications, with a flexible, extensible, and platform-agnostic architecture designed for both local and remote scenarios. ACP addresses integration overhead, limited compatibility, and developer lock-in by allowing agents that implement the protocol to work with any compatible editor, while editors that support ACP gain access to the broader ecosystem of ACP-compatible agents. Similar in spirit to how the Language Server Protocol standardized language server integration, ACP decouples agents and editors so both sides can innovate independently while developers choose the best tools for their workflow.Starting Price: Free -
5
rawctx
rawctx
rawctx is an AI answer evidence layer for teams shipping customer-facing AI assistants, copilots, and agents. It records each answer with the approved meaning reference, source/context references, model-run metadata, trace IDs, correction history, and exportable proof bundles. Teams can audit why an AI answer was shown, which evidence and business definition it used, what changed after review, and whether trust proof status is anchored, pending, or local-only. rawctx supports answer audit logs, JSON/CSV/proof exports, source_ref evidence binding, and public or private verification workflows for high-stakes support, sales, finance, legal, and operations use cases.Starting Price: $9.99/1000logs -
6
Graphify
Graphify
Graphify is an open source knowledge graph engine that turns any input, including code, docs, papers, meetings, images, browser tabs, and commits, into one traversable graph with complete recall. It is built as persistent memory for AI coding assistants, giving tools like Claude Code, Codex, OpenCode, Cursor, Gemini CLI, GitHub Copilot CLI, Aider, Factory Droid, Kimi Code, Kiro, Pi, and Google Antigravity a queryable understanding of a project instead of making them repeatedly grep through files. Users can point Graphify at any directory, and it builds an initial corpus through AST extraction, semantic analysis, and Leiden clustering, transforming an entire codebase or document corpus into a graph in one pass. Unlike RAG pipelines that re-embed everything on every change, Graphify maintains a living graph that updates only affected nodes and edges when files change, allowing the rest of the corpus to stay intact even at enterprise scale.Starting Price: Free -
7
Hindsight
Vectorize
Hindsight is an agent memory system built to create smarter AI agents that learn over time instead of starting every conversation from zero. Most agent memory systems focus on recalling conversation history, but Hindsight is focused on making agents learn, not just remember. It gives AI agents persistent long-term memory using biomimetic data structures, helping them retain facts, recall relevant context, and reflect on experience as part of reasoning. Hindsight is designed for agents that need to understand who a user is, what has been discussed, what preferences have emerged, what decisions were made, and how behavior should adapt across sessions. It provides three core operations: retain, recall, and reflect. Retain stores new information, recall retrieves the right memories when needed, and reflect helps agents synthesize observations, form mental models, and learn from prior interactions.Starting Price: Free -
8
Pipecat
Pipecat
Pipecat is an open source framework and ecosystem for building real-time voice and multimodal conversational AI agents. It gives developers everything they need to create, deploy, and scale AI applications that can see, hear, and speak, while orchestrating audio, video, AI services, transports, and conversation pipelines with ultra-low latency. The core Pipecat framework is a Python-based system for building voice and multimodal AI pipelines, helping teams connect components such as speech-to-text, LLMs, text-to-speech, vision, video, transports, and business logic without manually wiring every service from scratch. Pipecat is designed to be vendor-neutral and composable, supporting more than 100 AI services so developers can choose the models and providers that fit each use case. Its ecosystem includes Pipecat Subagents for coordinating specialized agents with handoff, task dispatch, and distributed deployment.Starting Price: Free -
9
Vision Agents
Stream
Vision Agents is an open source Python framework for building low-latency voice and video AI agents with any model. It lets developers plug in LLM, speech, and vision models from more than 25 providers and ship real-time agents for telehealth, voice support, live coaching, video analysis, interactive avatars, security monitoring, sports commentary, and other multimodal applications. It is designed to help teams build agents that can listen, speak, see, process media, call tools, and respond in real time while running on Stream’s global edge network with sub-500ms latency. Developers can build a first agent in minutes, using a small Python setup with Gemini Realtime, OpenAI, Deepgram, ElevenLabs, Stream, or other supported providers. Vision Agents supports both real-time speech-to-speech models and custom STT/LLM/TTS pipelines, giving teams either the fastest path to a working voice agent or full control over speech recognition, language reasoning, text-to-speech, etc.Starting Price: Free -
10
Docling
Docling
Docling is an easy-to-use, self-contained, MIT-licensed open source toolkit for converting messy documents into structured data and simplifying downstream document and AI processing. It can parse many popular document formats into a unified and richly structured Docling Document, including PDF, DOCX, PPTX, XLSX, HTML, Markdown, AsciiDoc, CSV, images, audio, and scanned pages through an OCR engine of the user’s choice. Docling detects tables, formulas, reading order, chunks, bounding boxes, page headers and footers, pictures, captions, code, list items, paragraphs, cells, and document structure, making extracted content easier to process, search, and ingest into AI, RAG, and agentic systems. It can export parsed documents to JSON, text, Markdown, HTML, and Doctags, giving developers flexible outputs for pipelines and applications. Docling stores and traverses components according to reading order, partitions documents into bite-sized contiguous text chunks.Starting Price: Free -
11
Manufact
Manufact
Manufact is a platform to build and deploy MCP apps and servers, giving teams a fast path to the ChatGPT Apps Store, Claude Connectors, and every surface where users and agents already work. The mcp-use SDK is the full-stack MCP framework to develop MCP apps for ChatGPT and Claude, as well as MCP servers for AI agents. Manufact covers every step of the MCP lifecycle with no extra tools: build from an SDK, a skill, or a vibe; deploy with one push; publish with marketplace checklists and generated submission assets; iterate with Cloud Inspector; and monitor with analytics, session replay, traces, error rates, and alerts. Teams can scaffold with the MCP-use SDK, install a skill into a coding agent, describe an app and watch it scaffold, or drop in an existing MCP server unchanged. Manufact Cloud connects to a repo once, then every push auto-deploys, with preview URLs for pull requests, custom domains, and SSL handled.Starting Price: $25 per month -
12
Ornith-1.0
DeepReinforce
Ornith-1.0 is a self-improving family of models built specially for agentic coding tasks. It spans the full spectrum from compact 9B Dense models suitable for edge device deployment to 397B MoE frontier-scale models optimized for maximum performance, with variants including 9B Dense, 31B Dense, 35B MoE, and 397B MoE. Built on top of pretrained Gemma 4 and Qwen 3.5, Ornith-1.0 achieves state-of-the-art performance among open-source models of comparable size on coding benchmarks. Its key innovation is a self-improving training framework that learns to generate both solution rollouts and the task-specific scaffolds that guide those rollouts. Instead of relying on fixed, human-designed harnesses, Ornith-1.0 treats the scaffold as a learnable object that co-evolves with the policy, allowing the model to jointly optimize the orchestration and the final solution.Starting Price: Free -
13
Sendmux
Sendmux
Sendmux gives AI agents their own inbox and a controlled way to send email. Each agent can receive messages, read threads, handle attachments, send replies, and react to new mail through signed webhooks or Server-Sent Events. Teams can send through Gmail OAuth, Outlook OAuth, SMTP providers, or managed Amazon SES. Sendmux can spread sending across providers with weighted routing, sender-domain routing, health checks, failover, and quotas by second, minute, hour, or day. Developers get Mailbox, Sending, and Management APIs, OpenAPI docs, SDKs, the sendmux CLI, MCP tools, and Sendmux Skills. The platform also covers tenant isolation, mailbox-scoped API keys, custom domains, logs, metrics, and usage-based pricing with no per-seat or per-mailbox fees.Starting Price: $0.15 per 1000 emails -
14
Open Voice OS
Open Voice OS
Open Voice OS is a community-driven, open source voice AI platform for creating custom voice-controlled interfaces across devices with NLP, a customizable UI, and a focus on privacy and security. Built by developers from Linux and FOSS communities around the globe, it is an open platform for making cutting-edge voice assistance technology available to everyone. Open Voice OS is multi-platform, designed for embedded headless devices, single-board computers with screens, DIY smart speakers, Raspberry Pi devices, Mark I and Mark II devices, Linux desktops, laptops, and Docker containers. As a voice operating system, OVOS goes beyond a traditional “Hey Mycroft…” assistant and provides the foundational tools and frameworks for integrating voice interaction into robots, automation systems, smart furniture, mirrors, cloud-based voice services, embedded devices, and smart TVs.Starting Price: Free -
15
Gemini Managed Agents
Google
Gemini Managed Agents provides a visual way to prototype and learn how to build managed agents without having to create and write API calls. Managed agents on the Gemini API give developers a configurable agent harness where a single API call provisions a secure Linux sandbox for the agent to reason, execute code, manage files, and browse the web autonomously. In AI Studio Playground, users can switch to the Agents tab and start from pre-built templates that configure the base Antigravity Agent with tools, environment settings, instructions, and skills. The Antigravity Agent is a general-purpose managed agent powered by Gemini 3.5 Flash, able to run code, manage files, search the web, and be extended with custom instructions, skills, and data. Agent behavior, persona, and capabilities are defined through files in the environment, including AGENTS.md for system instructions and persona, and SKILL.md files for specific capabilities and workflows.Starting Price: $1.50 per 1M tokens -
16
Engram
Weaviate
Engram is a fully managed memory and context service purpose-built to help AI agents remember, learn, and improve over time. Instead of treating memory as an ever-growing pile of raw conversations and events, it turns noisy interaction data into structured, durable, and evolving memories. Applications can send raw text, complete conversations, or pre-extracted facts through a REST API or Python SDK without preprocessing. Engram then runs asynchronous pipelines that extract relevant information, transform it by deduplicating and reconciling it with existing knowledge, and commit a clean memory state without blocking the application’s main workflow. It resolves inconsistencies, adapts to changing preferences and time-evolving facts, and keeps context relevant and efficient. Agents can retrieve ranked memories in real time through vector similarity, BM25 keyword search, or hybrid retrieval, reducing the need to resend entire conversation histories.Starting Price: $45 per month -
17
Gemini 3.5 Flash-Lite
Google
Gemini 3.5 Flash-Lite is Google’s fastest model in the Gemini 3.5 series, designed for low-latency tasks and high-throughput developer workflows such as agentic search, document processing, coding, and large-scale data analysis. It delivers 350 output tokens per second and significantly improves on previous Flash-Lite generations in both quality and agentic performance. Developers can configure its thinking level to match the workload: minimal or low thinking supports fast execution for high-volume tasks, while higher thinking levels enable more complex, multi-step subagent workflows. Built-in computer-use capabilities allow the model to interact reliably with digital environments across supported surfaces. Gemini 3.5 Flash-Lite also advances coding, long-context understanding, and real-world task execution, outperforming Gemini 3.1 Flash-Lite across key evaluations and even surpassing Gemini 3 Flash on several agentic and software-engineering benchmarks.Starting Price: $0.30 per 1M input tokens -
18
LM Studio Bionic
LM Studio
LM Studio Bionic is an AI agent built for getting real work done with open models across coding, research, documents, files, and general knowledge work. It gives users flexible control over where models run: locally on their device, through LM Link, or with frontier open-source models in LM Studio Secure Cloud for heavier tasks. Local models are powered by the LM Studio runtime and can be downloaded directly inside the app, while cloud requests use Zero Data Retention and are processed without being stored after completion. For coding, users can connect a local folder as a Code project and ask Bionic to inspect a codebase, explain unfamiliar logic, search for relevant files, trace behavior, edit code, or debug issues. Inline diffs make changes easy to review as the agent works. Work projects support documents, PDFs, presentations, spreadsheets, and local directories, allowing Bionic to generate new files, organize materials, summarize content, edit existing work, and more.Starting Price: Free -
19
Kastra
Kastra
Kastra is the authorization layer for AI systems, deciding what agents, models, and AI tools are allowed to do before they do it. It sits in the execution path of every prompt, tool call, shell command, database operation, and API request, evaluates each action against deterministic, attribute-based policy, and returns an allow, deny, redact, or escalate decision in under a millisecond. Unlike monitoring products that observe AI after it acts, Kastra blocks unauthorized behavior before it reaches a tool, API, database, or production system. Its unified control plane combines a policy engine, edge decision points, integrations, and a tamper-evident evidence vault that signs every decision for audit and replay. Kastra Edge brings local enforcement to developer machines, protecting Claude Code, Cursor, Codex CLI, and other coding agents from destructive commands, secret exfiltration, unsafe file writes, and unauthorized tool use.Starting Price: $19.99 per month -
20
Prefactor
Prefactor
Prefactor is a real-time evaluation, observability, and reliability platform for production AI agents. It scores every run the moment it happens for quality, drift, cost, and data risk, then wires those evaluations into action so a failing agent is caught live instead of only appearing on a dashboard afterward. Teams can observe every model call, tool invocation, and decision as structured traces and spans, run LLM-as-judge, technical, qualitative, and custom evaluations on each step, and attach context from GitHub, Linear, Jira, databases, internal APIs, or other sources as ground truth. When a run crosses a defined threshold, Prefactor can block or throttle it, pause a sensitive action, or route it to a person for approval, modification, or rejection before execution, with every decision logged. The CLI discovers agents without a platform migration, while TypeScript and Python SDKs provide native support for LangChain, Claude, Vercel AI, OpenClaw, and LiveKit.Starting Price: $250 per month -
21
Assistable
Assistable
Assistable is an AI conversation platform for building, testing, deploying, and observing hybrid AI agents that answer calls, texts, WhatsApp messages, and web chats with one persistent memory across every channel. Agents can handle full customer conversations, understand intent, pull information, qualify leads, book, reschedule, or cancel appointments, create tickets, update contacts, add notes and tasks, and escalate to humans when configured to do so. Because memory carries across channels, a customer can call first and continue later by text or chat without repeating the conversation. Users can describe what an agent should do in plain English, connect Assistable through MCP, and have the system create the agent, attach tools, connect channels, and assign a phone number. Agents can read live calendar availability, follow up with leads, and support both inbound and outbound conversations.Starting Price: $225 per month -
22
Waterr
Waterr
Waterr is an AI meeting platform that lets teams delegate live conversations to autonomous AI personas that can see, hear, and speak with participants. Users describe what a meeting is for in plain English or start from a ready-made template, with scenarios for candidate screening, sales discovery, customer onboarding, coaching, feedback interviews, support, research, and recurring operational workflows. Each persona can have its own face, real-time voice, background, and conversation behavior, creating an experience designed to feel more like a meeting with a person than a chatbot. Teams can preview the persona, scenario prompt, and full conversation flow, then test it instantly in a real meeting before sharing it with participants. The same scenario can also be embedded on a website as an avatar, inline widget, or talk-to-us button. For developers, Waterr provides one API for creating AI meetings, with authentication, SDKs, webhooks, rate limits, and support for custom avatars.Starting Price: $29 per month -
23
Kilo Gateway
Kilo
Kilo Gateway is a universal AI inference gateway that routes LLM requests to any provider through one standardized endpoint, giving developers access to hundreds of hosted and open models without rewriting their applications for each provider. It provides unified access to models from Anthropic, OpenAI, Mistral, and other providers, while also supporting bring-your-own-key configurations that let teams connect existing provider credentials through centralized infrastructure. The gateway is compatible with standard AI SDKs, making it possible to switch providers while keeping the same integration surface. Its infrastructure handles routing complexity and load balancing across direct providers and external gateways to improve availability and resilience. Auto Model can route each request to the best available model while keeping routing decisions, model behavior, and usage visible and controllable.Starting Price: $19 per month -
24
Mireye
Mireye
Mireye is infrastructure for physical-world AI agents, providing one API for data, enrichment, tools, and signals. Agents such as Claude, ChatGPT, Gemini, Cursor, Grok, Copilot, DeepSeek, or custom systems can ask questions about places through one API with MCP tools and receive cited data indexed from the physical world. Mireye can resolve an address, fetch location-specific fields, measure distances, and return the plan it executed so results can be replayed. Its geocoding endpoint converts U.S. street addresses into coordinates while reporting how each location was derived and refusing low-confidence matches rather than guessing. Lookup expands a point into place context such as county, census tract, congressional district, metro area, timezone, elevation, and flood status. Fetch returns typed values for requested fields or presets at U.S. locations, with a source, fetch time, and confidence attached to every value.Starting Price: $19 per month -
25
Isle
Isle
Isle provides managed desktop application environments for computer-use agents, with guardrails, application monitoring, and artifact checkpoints built in. Instead of giving an agent a generic machine and leaving it to manage the application itself, Isle wraps each application in the monitoring, containment, and recovery controls needed to complete real work. Environments launch with the intended software already installed, configured, and open, while health checks verify that the application itself is ready rather than simply confirming that the underlying machine is online. Developers can connect any computer-use model or agent loop using familiar screenshot, mouse, keyboard, file-transfer, and execution primitives. Artifact checkpoints protect useful progress before important actions and can be restored if something goes wrong, while recovery controls allow agents to restart applications and return to the latest safe state.Starting Price: $29 per month -
26
Holo4
H Company
Holo4 is a family of generalist agentic AI models from H Company designed to operate software across graphical interfaces, code environments, MCP servers, APIs, desktop applications, the web, and mobile devices. The family includes a 27B dense model and a 35B-A3B Mixture-of-Experts model with 3B active parameters, both supporting a 256K context window. Unlike agents optimized around a single interface, Holo4 can click and type on screens, write and execute code, and call MCP or API tools depending on what a task requires. The models were trained with supervised fine-tuning and reinforcement learning on agentic workflows spanning desktop, web, MCP/API, mobile, coding, multimodal reasoning, and GUI grounding. Holo4 27B achieved 85.2% on OSWorld at a reported cost of $0.08 per task, while Holo4 35B-A3B scored 80.8% at $0.05 per task in H Company's evaluation.Starting Price: $0.40 per 1M tokens (input) -
27
Laya AI
Laya AI
Laya AI is an open source family of non-autoregressive System 1 decision models built for fast, typed decisions instead of free-form text generation. Given a state and structured questions, Laya evaluates them in a single forward pass and returns outputs that application code can use directly. It supports three decision types: choice, which selects among named options and returns probabilities; score, which places an input on an ordered rubric such as urgency or severity; and noul, which estimates the probability that a proposition is true. Laya is designed for workloads such as agent and model routing, support ticket triage, guardrails, intent classification, risk and fraud scoring, workflow escalation, RAG filtering, moderation, and phishing detection. It can route requests by department, score urgency, flag refund or churn signals, detect policy or abuse conditions, and escalate uncertain cases to a larger model or human reviewer.Starting Price: Free -
28
Tev1
Together AI
Tev1 is a fast decision and classification model built by Together AI and fine-tuned from Qwen3.5 4B. It is designed around a simple interface: provide a piece of state, a question, and 2 to 24 labeled options, and the model returns exactly one letter that software can map back to a semantic key. This structure supports production tasks such as routing support tickets, evaluating automated returns, applying policy rules, rating sentiment, categorizing documents, and inserting deterministic decision points into larger agent pipelines. The model has 4.7B parameters, a 32.8K-token context window, and runs with non-thinking generation for fast, predictable classification. It was trained on 37,840 examples spanning intent detection, boolean comprehension, sentiment, policy application, routing, news categories, and research-paper taxonomy. Recommended settings use temperature 0 and a very small output budget so responses remain stable and easy to parse.Starting Price: $0.04 per 1M tokens -
29
Span-01
Respan
Span-01 is a hyper-parallel reasoning classifier built to detect complex behaviors across long production traces in a single forward pass. Rather than generating tokens sequentially, it reasons over natural-language behavior definitions and returns direct probabilities for present, absent, and not_observable. The model is trained for general classification reasoning with RLAIF before being specialized for behavior detection, allowing it to apply previously unseen definitions to new behaviors and domains instead of relying on a fixed taxonomy. At inference time, hyper-parallel definition branches evaluate multiple behaviors against the same trace context simultaneously, while hybrid attention helps preserve reasoning over long, complex inputs. Span-01 is designed for production monitoring across safety, reliability, grounding, agent behavior, and response quality, including jailbreak and prompt injection detection, privacy, and secrets.Starting Price: $199 per month -
30
boat
boat
boat provides powerful cloud virtual machines and persistent sandboxes built for long-running AI agents, allowing users to spawn thousands of cloud computers with browsers and persistent files without managing the entire infrastructure themselves. Each sandbox is a full Ubuntu Linux virtual machine with SSH and SCP access, Docker support, dedicated IPv4 or IPv6 addresses, a 60 fps virtual desktop, and fast disk-level forking from snapshots. Docker, VS Code, Chrome, Ghostty, GitHub CLI, Rust, Node.js, and Bun come pre-installed, enabling users to run any development tool, language, or workflow. Boat supports agents such as Claude Code, Codex, and OpenCode, with tools to prompt, observe, and control agent environments remotely through its CLI, HTTP API, and Python or TypeScript SDKs. Users can execute commands, copy files, monitor agent messages and tool calls, host applications through public HTTPS URLs, and access integrated desktops and browsers.Starting Price: $20 per month
