Top AI agents, ranked and verified
25 agents scored out of 100 on 8 weighted criteria. Prices come from each vendor's own page and were checked on ; 21 of 25 prices are verified, the rest are marked.
| # | Agent | Category | Score | Pricing (USD unless noted) |
|---|---|---|---|---|
| 1 | ChatGPT Agent OpenAI Browses, runs code and works with files from one prompt, inside the most used AI app. Top on reach and access: it sits inside the most used AI app, with public plan prices. Score breakdown
| General | 90/100 | Plus $20/mo; Pro from $100/mo source, 2026-10-31 |
| 2 | GitHub Copilot (coding agent) Microsoft / GitHub Cloud coding agent and code review, the safest enterprise default for many teams. Best governance and reliability score among coding agents for teams; public per-seat pricing. Score breakdown
| Coding | 90/100 | Pro $10/mo; Pro+ $39/mo; Max $100/mo; Business $19/seat; Enterprise $39/seat. Coding agent on paid plans source, 2026-10-31 |
| 3 | Claude Code Anthropic Terminal-native coding agent with a permission-gated execution loop. Highest autonomy among coding tools, with permission gates and public pricing; momentum is strong. Score breakdown
| Coding | 87/100 | Pro $20/mo ($17 billed annually); Max $100/mo; Ultra $200/mo source, 2026-10-31 |
| 4 | Microsoft 365 Copilot Microsoft Agents inside Word, Excel, Teams and Outlook; Researcher and Analyst are pre-built. Top on governance and integrations inside Microsoft 365; access is limited to licensed organizations. Score breakdown
| Productivity | 86/100 | Copilot Business add-on $21/user/mo ($18 promo through Dec 31, 2026, annual commitment) source, 2026-10-31 |
| 5 | Gemini (Deep Research, Agent Mode) Google Cited deep research and agentic browsing tied into Google Workspace. Huge reach and a free tier, but agent mode is named only for the top plan, so autonomy is capped. Score breakdown
| General | 85/100 | Free tier; AI Plus $4.99/mo; AI Pro $19.99/mo; AI Ultra from $99.99/mo. Agent Mode is named only for Ultra source, 2026-10-31 |
| 6 | OpenAI Codex OpenAI Delegates coding tasks to cloud and local agents inside ChatGPT plans. Strong autonomy and momentum, bundled into plans many teams already pay for. Score breakdown
| Coding | 84/100 | Included in Plus $20/mo; Pro from $100/mo; Business $20/user/mo billed annually source, 2026-10-31 |
| 7 | Microsoft Copilot Studio Microsoft Low-code builder for custom agents across Microsoft 365 and Power Platform. Excellent governance and integrations for building agents in Microsoft environments; metered pricing. Score breakdown
| Enterprise | 84/100 | Pay-as-you-go $0.01 per credit; capacity pack $200/mo for 25,000 credits (annual billing) source, 2026-10-31 |
| 8 | Salesforce Agentforce Salesforce Agents grounded in CRM data for sales, service and operations. Strong governance and reliability on CRM data; access is limited to Salesforce customers. Score breakdown
| Enterprise | 83/100 | $2 per conversation; Flex Credits $500 per 100k; employee add-ons $125/user/mo source, 2026-10-31 |
| 9 | Claude (app, agentic tasks) Anthropic Long-running work on files, documents and code with tool use. Balanced across criteria; long-running file and code work with a free tier and clear plans. Score breakdown
| General | 82/100 | Free; Pro $20/mo; Max $100/mo; Team from $20/seat/mo billed annually source, 2026-10-31 |
| 10 | Zapier Agents Zapier Agents on top of the widest app catalog in automation. Widest app catalog and clear pricing; autonomy and momentum are moderate. Score breakdown
| Automation | 79/100 | Free 400 activities/mo; Pro $33.33/mo billed annually (1,500 activities) source, 2026-10-31 |
| 11 | Cursor Anysphere AI-first IDE with a multi-file agent; the most cited daily driver for developers. Strong daily-driver for developers; governance is the weak spot, and plans have been renamed recently. Score breakdown
| Coding | 78/100 | Free; Plus $8/mo; Premium $30/mo; Super $100/mo; Ultra $200/mo; Teams from $30/user source, 2026-10-31 |
| 12 | Gemini Enterprise / Vertex AI Agent Builder Google Agents grounded in company data with multi-agent orchestration. Strong governance on company data; adoption and momentum are still catching up. Score breakdown
| Enterprise | 76/100 | Business $21/seat/mo; Standard/Plus $30/seat/mo source, 2026-10-31 |
| 13 | n8n n8n Open-source workflow platform with agents; can be self-hosted. Open source and self-hostable with transparent pricing; autonomy and adoption are moderate. Score breakdown
| Automation | 75/100 | Cloud Starter €20/mo and Pro €50/mo billed annually; Community Edition free to self-host source, 2026-10-31 |
| 14 | Amazon Bedrock AgentCore Amazon (AWS) Runs and governs agents inside AWS. Strong governance on AWS with fully published unit prices; adoption is still developing. Score breakdown
| Infrastructure | 75/100 | Usage-based: Runtime $0.0895 per vCPU-hour; Gateway $0.005 per 1,000 calls; up to $200 credits for new AWS customers source, 2026-10-31 |
| 15 | Fin Intercom One of the most cited customer-support agents; priced per outcome. Reliable support agent with a transparent per-outcome price; narrow in scope. Score breakdown
| Customer support | 74/100 | From $0.99 per outcome; no seats for the standalone product source, 2026-10-31 |
| 16 | Devin Cognition Autonomous software engineer that plans, codes, debugs and opens pull requests. Very autonomous software engineer; reliability and breadth are lower than the leaders. Score breakdown
| Coding | 72/100 | Free; Pro $20/mo; Max $200/mo; Teams usage-based, $80/mo minimum source, 2026-10-31 |
| 17 | UiPath AI Agents UiPath RPA plus agents for companies that already automate with UiPath. Reliable, governed automation for existing UiPath customers; enterprise tiers are on request. Score breakdown
| Enterprise | 72/100 | Community free; Basic from $25/mo; Standard and Enterprise on request source, 2026-10-31 |
| 18 | Replit Agent Replit Lets non-engineers ship deployed full-stack apps. Lets non-engineers ship apps and has clear pricing; reliability and governance are middling. Score breakdown
| Coding | 71/100 | Core $20/mo ($18 annual); Pro $100/mo ($90 annual) source, 2026-10-31 |
| 19 | Perplexity (Comet) Perplexity Research agent with structured citations, plus an agentic browser. Best-in-class citations and easy access; fewer agent controls, and its prices are not confirmed here. Score breakdown
| Research | 70/100 | See the official pricing page unverified, official page |
| 20 | Manus Manus AI Autonomous multi-step agent often cited for end-to-end task completion. High autonomy for end-to-end tasks; weaker on governance, and we could not confirm its prices. Score breakdown
| General | 68/100 | Free, Pro and Team plans are credit-based; the official docs page shows no prices unverified, official page |
| 21 | ServiceNow AI Agents ServiceNow Agents for IT and HR workflows in large companies. Reliable and well governed for IT and HR; no public prices and access is via sales only. Score breakdown
| Enterprise | 68/100 | No public prices; quote or demo only source, 2026-10-31 |
| 22 | Google Antigravity Google Multi-agent development platform with parallel agents on Gemini models. Interesting parallel-agent design, free for individuals; paid tiers are not itemized. Score breakdown
| Coding | 67/100 | Free for individuals; paid tiers are not itemized on the official page source, 2026-10-31 |
| 23 | Glean Glean Search and agents across all of a company's apps. Excellent integrations and governance; no public prices and access is via a demo. Score breakdown
| Enterprise | 67/100 | No public prices; quote or demo only source, 2026-10-31 |
| 24 | ChatGPT Atlas OpenAI Browser with a built-in agent. Agent built into a browser; governance is thin and availability details are not confirmed here. Score breakdown
| Browser | 60/100 | See the official page unverified, official page |
| 25 | Grok xAI Real-time access to X data, useful for markets and news. Unique real-time X data; weaker on governance and integrations, and prices are not confirmed here. Score breakdown
| General | 58/100 | See the official page unverified, official page |
How the scores work
Each agent gets 1 to 5 on 8 criteria. A score converts to points in proportion to the criterion's weight, and the points add up to a total out of 100. The scores are Deyron Labs editorial judgments based on public rankings and the vendors' own pages; they are not benchmarks. Ties keep the order of our data file.
| Criterion | Weight | What it measures |
|---|---|---|
| Adoption | 20 | Reach and real usage: users, customers, how often it is cited in developer and business surveys. |
| Autonomy | 15 | How far it gets on a multi-step task without help. |
| Reliability | 15 | Quality of results and stability on real work. |
| Security & governance | 10 | Permissions, admin controls, audit and compliance options. |
| Integrations | 10 | Connectors, app catalog, APIs and developer ecosystem. |
| Ease of access | 10 | Free tier, entry price and effort to start. |
| Pricing transparency | 10 | Prices public and itemized on the vendor page. Neutral (3) when we could not verify the page. |
| Momentum | 10 | Pace of releases and growth in 2026. |
Links go to company domains and official pricing pages. Prices change often: every row shows the date it was checked, and rows we could not confirm say so instead of guessing.
Spotted a wrong price or a score you disagree with? Suggest a correction and include the official page that shows it.