Skip to contentT/TTokens or Towers

Compare

LeaderboardEvery tracked model, ranked and priced side by sideBenchmarksWhat each test measures and who leads itPricesAPI, media, and hardware rates with a cost calculator

Plan

Stack plannerSix questions to a priced local and hosted stackShortlistOur picks for subscriptions, APIs, and hardwareEnterpriseSeat plans against self-hosting for a whole team

Build

HardwareGPUs and workstations, and which models they fitHostsWhere to rent open-weight inference, and for how muchHarnessesCoding agents and the models they drive

Lab

FindingsResults from our own evaluationsScanAudit a repo’s AI usage in the browserMethodWhere the numbers come from and how we check them
Data checked September 25, 2026
Data checked September 25, 2026
Plan a stack
T/TTokens or Towers

An independent guide to AI models, what they cost, and the hardware to run them. No account, no API key, no sponsors.

Compare

  • Leaderboard
  • Benchmarks
  • Prices

Plan

  • Stack planner
  • Shortlist
  • Enterprise

Build

  • Hardware
  • Hosts
  • Harnesses

Lab

  • Findings
  • Scan
  • Method
Data checked September 25, 2026 · Every number links to its sourceRSS · Source on GitHub

Enterprise

Seats for the team, or a tower for the team.

Price team plans per developer, multiply by headcount, and compare them with shared hardware running open models. Every number cites a source, and the caveats stay next to the answer.

Pick a seat count and workload mix. The three routes recompute as you change the inputs.

Team inputs

Set the seats and how the team works.

The three routes recompute as you change these inputs.

Accepted range is 2 to 500. Values outside that clamp on blur.

Workloads
Data stance
Volume per developer

Three team routes

The lowest bill at this headcount.

Monthly totals scale with 20 seats at regular volume per developer. The lowest mid-range total gets the accent treatment.

Seat plans

AnthropicTeam seat plan

Standard seats include Opus 5.5, Sonnet 5.5, Haiku 4.5, and Claude Code.

Monthly total
$400
Per seat
$20
What this is
  • Claude Team Standard is the lowest public team rate that covers 20 seats.
  • The estimate uses the published per-seat rate for every developer.
Claude plan pricing ↗
Reality check
  • The $20 monthly equivalent requires annual billing ($25 month to month), for teams of 2 to 150. Premium seats cost $100 ($125 month to month) for five times Standard usage. Self-serve Claude Enterprise costs $20 per seat plus usage at API rates.
  • Negotiated enterprise pricing usually lands below public list rates at larger seat counts.
Cheapest

Metered API

Monthly total
$280–$460
Per seat
$14–$23
What this is
  • Metered APIs route each selected workload through the lowest-cost covering models in the catalog.
  • The estimate applies 100 million tokens to every developer, so each seat adds the same usage range.
Reality check
  • The range assumes 70 percent input tokens and 30 percent output tokens. Prompt caching discounts are excluded.
On prem

On-prem serving

Monthly total
$556
Per seat
$28
Capex
~$20,000
Machines4 × DGX Spark128 GB unified · 273 GB/s

Prioritizes model capacity and privacy over raw token speed. The $4,699 Founders Edition is out of stock, so this uses the cheapest available unit.

What this is
  • DGX Spark capacity appliance is the cheapest catalog option at the top seat capacity that covers the selected workloads locally.
  • One box serves about 6 developers under an interactive coding load. Hardware cost is spread over 36 months.
Reality check
  • Serving and operations labor are excluded from the math. Electricity is also excluded.
  • Capacity assumptions need validation against your real workloads.

Need the one-person version? Open the solo planner for individual stack math.