CompareCohere North

Fig vs. Cohere North

Frontier models today, or your own models behind your own wall.

Cohere North is the enterprise AI workspace for organizations that can't let data leave the building: agents and automations that run on two GPUs in your VPC, on-prem, or fully air-gapped, on Cohere's own Command models. Fig is the other answer to the same question — a managed platform your team can use this afternoon, routing across frontier models from multiple labs, in Slack and on every device, priced on usage.

Fig vs. Cohere North, in short

Cohere North is built for sovereignty: it runs inside your own infrastructure — VPC, on-prem, or air-gapped — on Cohere's Command models, sold through an enterprise sales process with reference deployments measured in months. Fig is a managed platform available today that routes across frontier models from multiple labs, runs in Slack and on every device, and bills usage with no seats or minimums. Choose on where your data must live and which models your people should get.

Trusted by people from

AmazonAppleU.S. NavyTeliaTraneMerckKeller WilliamsUC System

Fig vs. Cohere North at a glance

FeatureFigCohere North
Getting started
Self-serve today. No seat minimum, no sales cycle to begin.
Sales-led; reference deployment reached production in 90 days.
Pricing model
Usage only. No seats; unlimited members; one shared credit pool.
Custom; five-figure+ annual minimums estimated (third-party).
Models
Frontier models from multiple labs, routed per task.
Command models by default; external LLMs available in Automations.
Where it runs
Slack (in-channel), desktop, mobile, and web.
Web workspace and API; no documented mobile or desktop app.
Connectors & actions
Native connectors, computer use, MCP, approval gates on actions.
Named connectors, any MCP server, approval checkpoints.
Company context
Enterprise Graph: people, teams, projects, activity; follows source permissions.
Grounded, cited answers; access enforced at the data source.
Agents
Agent Builder, Orchestration, Governance — on the same usage pool.
Agent Studio and North Automations (GA July 2026).
Admin & compliance
SSO (SAML/OIDC), SCIM 2.0, scoped roles, audit export, break-glass re-auth.
SOC 2 Type II, ISO 27001/42001 (corporate); FedRAMP High path.
Deployment
Managed cloud only; no on-prem or air-gapped option.
VPC, on-prem, air-gapped, sovereign cloud; two GPUs minimum.

Included   Partial or with conditions   Not offered

$0per seat, at any team size
Unlimitedmembers on one shared usage pool
Every labfrontier models, routed per task

Pricing: Fig vs. Cohere North

Fig

No seats. Team members are free to add, and usage is billed from your team's shared credit pool — you only pay for what the team actually consumes, starting today.

See how usage pricing works →

Cohere North

Custom and unpublished. Third-party analyses describe a hybrid of usage fees plus fixed licensing, starting in the five figures a year with committed minimums and reaching multi-six-figures when dedicated inference is included. Cohere's only published number is Model Vault dedicated inference at $4–10 per hour per instance. Treat the estimates as estimates.

North is priced for $1B-revenue enterprises in regulated industries — Cohere's own market research says so. Fig is priced so a single team can start without a contract.

Fig vs. Cohere North in detail

Sovereignty is the product

Cohere's bet is that a large share of enterprise AI will run inside the customer's own walls — and roughly 85% of its revenue comes from private deployments. North runs on two GPUs in your VPC, your data center, or an air-gapped room, on models Cohere trained and, in Command A+'s case, released as open weights. Fig is a managed platform and solves a different problem: giving a team the strongest frontier models available, today, without building or hosting anything.

The model behind the workspace

North defaults to Command, Cohere's enterprise model, with external LLMs available per step in Automations. Fig routes each task to the best frontier model for it, across multiple labs, and lets admins decide per team which models are in play — so your people get the strongest model available for the work in front of them, with no single lab's roadmap setting the ceiling.

Ninety days, or this afternoon

Cohere's flagship reference customer went from engagement to production in 90 days with an on-site workshop — fast for an enterprise deployment, and still a quarter. There's no trial and no self-serve path. Fig's first value is the first question a team asks in Slack; enterprise deployment and enablement layer on from there rather than gating the start.

Which is right for you?

Choose Fig if

Teams and mid-market companies that want frontier models from multiple labs, company context, and governed agents in Slack and on every device, available today and priced on usage — and whose data can live in a managed cloud platform.

Choose Cohere North if

Large regulated or public-sector organizations — Cohere targets $1B+ revenue — that must run AI inside their own VPC, data center, or air-gapped environment, prefer an open-weights model they control, and have the budget and timeline for a sales-led enterprise deployment.

Frequently asked questions

Is Fig an alternative to Cohere North?

For most companies, yes: both provide an AI workspace with agents, connectors, and grounded answers over company data. North is built for organizations that must deploy inside their own infrastructure on Cohere's models, sold through an enterprise sales process. Fig is a managed platform available today, routing across frontier models from multiple labs, billed on usage. If your requirement is air-gapped or sovereign deployment, North is built for that.

Does Cohere North publish pricing?

No. North is quoted as custom enterprise pricing. Third-party analyses estimate five-figure annual minimums rising to multi-six-figure deals with dedicated inference; the only published Cohere number is Model Vault at $4–10 per hour per instance. Fig has no seats — team members are free to add and usage is billed from a shared credit pool.

Which models does each platform use?

North defaults to Cohere's Command models — deployable on two GPUs, with Command A+ available as open weights — and its Automations can call external LLMs per step. Fig routes each task across frontier models from multiple labs, with per-team controls over which models are available.

Can Fig be deployed on-premises or air-gapped?

No. Fig is a managed cloud platform. Cohere North offers VPC, on-premises, air-gapped, hybrid, and Cohere-managed dedicated deployment, and never routes customer data through Cohere's systems.

How long does each take to get running?

Fig is self-serve — a team is working in Slack, desktop, mobile, or web the same day. Cohere North is sales-led with no trial; Cohere's own reference customer reached production in 90 days following an on-site workshop.

See Fig on your team's real work.

Create a team, connect your tools, and ask your first question in Slack today — or walk through it with us.

Compare Fig with other platforms →

Be the next big thing

Dream big, build fast, and grow far with Fig

Fig Desktop Coming Soon!