Nori vs Devin: two AI engineers, one runs on your own subscription (October 2026)

Which is better, Nori or Devin?

It depends on what you need around the agent, because both now ship an AI engineer: Devin from Cognition, Jiro from Nori. Pick Devin if you want its broad built-in automations, mature memory, Outposts below Enterprise, and a $20 entry plan, and you accept quota-plus-credits billing with Devin as your only harness. Pick Nori if you want Jiro, or any harness such as Claude Code, Codex, or Gemini CLI, on your own Claude or ChatGPT subscription, keys, or gateway, with flat plans and no token charges.

This page compares the two head to head, as of October 2026, using each vendor's own documentation. It is part of our comparison of cloud coding agents; see also Nori vs Factory and Nori vs Claude Tag.

What is Devin?

Devin is an autonomous coding agent from Cognition. You give it a task from the web app, Slack, an issue tracker, or an automation, and it works in its own cloud machine and opens a pull request for review. Cognition also ships Devin Desktop; according to Cognition's own blog, that is the product formerly called Windsurf. Devin is built for engineering teams that want a single managed agent rather than a platform for running several.

As of October 2026, Devin's documentation describes a broad product around that agent: automations that start sessions from Slack, GitHub, GitLab, Linear, Jira, PagerDuty, schedules, and webhooks; memory (“dreaming”), playbooks, knowledge, skills, MCP, and hooks; and integrations with GitHub, GitLab, Bitbucket, Azure DevOps, Slack, Microsoft Teams, Jira, and Linear. The model side is closed: Devin runs Devin, and its deployment docs say it “does not currently support third-party LLM API keys.”

What is Nori?

Nori ships Jiro, its own AI engineering teammate, and also runs any harness you prefer. Each session is a cloud machine where you run Jiro or another harness — Claude Code, Codex, Gemini CLI, Cursor, OpenCode, pi, or an ACP agent such as Droid, Goose, Amp, or Copilot CLI — on your own Claude or ChatGPT subscription, API keys, or LLM gateway. You drive it from Slack, Discord, the web UI, or the CLI, and you schedule work with cron and webhook automations. Teams have merged 100,000+ PRs with it.

Three things shape the product. We never charge for tokens: plans are per runtime, with no metering, resale, or markup, so we have no incentive to make your agent burn more and no reason to lock you to one model provider. Everything is swappable: harness, subscription, keys, gateway, and on Enterprise your compute, IdP, token provider, and memory provider. And the defaults are opinionated: a workflow skillset for plans, TDD, worktrees, and review, plus org skillsets, memories, automations, and checkpoints, so an agent starts each session knowing how your team works.

Meet Jiro

Jiro is Nori's own AI engineering teammate, and the closest Nori counterpart to Devin. It runs on Claude Code or Codex under the hood, on your own Claude or ChatGPT subscription or API key, so there are no token charges. It follows an opinionated engineering workflow: planning, test-driven development, review, and worktrees. You work with it in Slack or the browser, and on its first day it reads your repos, asks the team how you ship, and sets up the workspace. See jiro.bot.

Nori vs Devin, head to head

Enterprise deployment: where does the control plane run?

The control plane is the part that orchestrates sessions, holds transcripts and org state, and runs the agent loop. Devin's deployment docs say the agent “always resides within Cognition's Cloud”: Dedicated is a Cognition-hosted single-tenant VPC, and Outposts move only session machines onto infrastructure you control. Nori Enterprise deploys the full stack — control plane, agents, and data — into your own AWS VPC, Azure VNet, or on-prem environment. On hosted plans, Nori gives each organization a dedicated, single-tenant control plane and data store, and every session runs on its own isolated machine.

When to choose Devin

• You want the agent and the model bill from one vendor. Devin bundles agent, models, and machine time into Cognition's billing, so nobody manages model credentials. Jiro, Nori's agent, needs a Claude or ChatGPT subscription or API key that you bring.
• You need automation and integration breadth today. Devin documents triggers from Jira and PagerDuty and integrations with Bitbucket, Azure DevOps, and Microsoft Teams. Nori's documented surfaces are Slack, Discord, the web UI, the CLI, and cron and webhook automations; anything else goes through a webhook.
• You want your own machines without an Enterprise contract. Devin's Outposts run session machines on infrastructure you control on Pro, Max, and Teams plans. On Nori, bringing your own compute is an Enterprise feature.
• You want self-serve SAML and SCIM. Devin documents Okta, Entra ID, generic SAML 2.0 or OIDC, and SCIM. Nori offers SAML, SCIM, and OIDC on Enterprise.
• You want a $20 entry point. Devin Pro is $20/mo, and there is a free tier. Nori's paid plans start at $99/mo after a 30-day trial.
• You value built-in memory. Devin's memory, playbooks, and knowledge are mature, first-party features.

When to choose Nori

• You want an AI engineer without token billing. Jiro is a ready-to-work teammate in Slack or the browser that plans, works test-first, and reviews its changes, on the subscription you already pay for.
• Your team already has an agent it trusts. Nori also runs Claude Code, Codex, Gemini CLI, and the rest as-is, so the prompts, skills, and habits your engineers built locally carry over.
• You already pay for Claude or ChatGPT. Nori uses those subscriptions directly. Devin has no Claude subscription support, and a ChatGPT plan covers only GPT usage there.
• You want a predictable bill. Devin's usage model is token quotas and then prepaid credits. Nori's plans are flat, and we never charge for tokens, so heavier use does not change what you pay us.
• You route models through a gateway. Nori works with OpenRouter, Fireworks, Cloudflare, or any OpenAI- or Anthropic-compatible endpoint. Devin does not accept third-party keys.
• You need the whole stack in your environment. On Devin, the agent always runs in Cognition's cloud, even with Outposts or Dedicated. Nori Enterprise runs in your cloud or on-prem, with your compute, IdP, LLM gateway, token provider, and memory provider.
• Your team uses Discord. Nori is driven from Discord as well as Slack.

How to switch from Devin to Nori

1. List what Devin does for you today. Write down each automation (its trigger and prompt), each playbook, and the knowledge entries you rely on.
2. Start the free trial. Thirty days, five runtimes, no card. Pick the harness you want to standardize on.
3. Connect models. Sign in with your Claude or ChatGPT subscription, or add API keys or your LLM gateway.
4. Move playbooks into skills. Devin playbooks map well to skills in your org skillset; standing facts about your codebase become memories.
5. Recreate automations. Schedules become cron automations. Event triggers become webhook automations pointed at the event source.
6. Connect Slack or Discord and run both tools on the same few tickets for a week before you cut over.

Plan details are on our pricing section.

Frequently asked questions

Does Nori have its own agent?

Yes. Nori ships Jiro, its own AI engineering teammate (jiro.bot). Jiro runs on Claude Code or Codex under the hood with your own Claude or ChatGPT subscription or API key, so Nori never charges for its tokens. It works in Slack or the browser and follows a planning, test-driven development, and review workflow. Nori also runs any other harness, including Claude Code, Codex, Gemini CLI, Cursor, OpenCode, and ACP agents.

Can I use my Claude subscription with Devin?

No. As of October 2026, Devin's billing documentation lists no Claude subscription support. A ChatGPT Go, Plus, or Pro plan can pay for GPT usage in Devin, but not on the Free or Enterprise plans, and Devin quota still pays for machine time and non-GPT models. Nori lets you sign in with your own Claude or ChatGPT subscription.

Does Devin support bring your own API key?

Not as of October 2026. Devin's enterprise deployment documentation says it does not currently support third-party LLM API keys. Nori accepts your own API keys or an LLM gateway such as OpenRouter, Fireworks, Cloudflare, or any OpenAI- or Anthropic-compatible endpoint.

How much does Devin cost compared with Nori?

As of October 2026, Devin has a Free plan, Pro at $20/mo, Max at $200/mo, Teams from $80/mo, and Enterprise through sales, with usage drawn from token quotas and then prepaid on-demand credits. Nori has a free 30-day trial, Developer at $99/mo, Team at $250/mo, and custom Enterprise pricing, and never charges for tokens.

Can Devin run on my own infrastructure?

Partly. Devin's Outposts run session machines on infrastructure you control, on Pro, Max, and Teams plans, but the agent itself always runs in Cognition's cloud; Dedicated is a Cognition-hosted single-tenant VPC. Nori Enterprise runs in your cloud or on-prem with your own compute, IdP, LLM gateway, token provider, and memory provider.

Can I run Claude Code or Codex inside Devin?

No. Devin runs only Devin. Its handoff feature sends a task from Claude Code or Codex to a Devin session, rather than running those harnesses. Nori runs Claude Code, Codex, Gemini CLI, Cursor, OpenCode, and ACP agents directly.

Sources

Competitor facts were read from these vendor pages on October 19, 2026. Vendors change plans and features often; check the source before you decide.

• Devin docs — Self-serve billing (plans, quotas, on-demand credits)
• Devin docs — Enterprise deployment overview
• Devin docs — Paying with a ChatGPT plan
• Devin docs — Devin handoff
• Devin docs — Outposts overview
• Devin docs — SCIM and SSO
• Devin docs — Automations
• Devin docs — Memory
• Devin docs — Integrations overview
• Cognition blog — One year of building together (Windsurf and Devin Desktop)

Nori ships Jiro, its own AI engineering teammate, and runs any coding agent you prefer in the cloud — on your subscription or keys, from Slack, Discord, the web, or the CLI, for a flat price. Learn more at noriagentic.com, or join our community Slack.