Stop guessing
which AI plan
is actually cheaper.
// AI subscriptions hide real cost behind “requests” and “compute hours”. TokenLens normalises everything to cost/1M tokens — so the numbers speak for themselves.
// data verified 2026-03-04
Platform comparison
Normalised cost across all plans. Cost/1M tokens uses a 65% input / 35% output split.
| Platform | Plan | Price/mo | Token cap/mo | Cost/1M tokens | Models | Confidence |
|---|---|---|---|---|---|---|
| Cursor | Cursor Free | Free | 4.5M | $7.20 | claude-sonnetgpt-4o | est |
| Cursor | Cursor Pro | $20 | 20.3M | $7.20 | claude-sonnetgpt-4ogpt-4.1 | est |
| Cursor | Cursor Ultra | $200 | 202.5M | $12.00 | claude-opusclaude-sonnetgpt-4.1gpt-4o | est |
| Windsurf | Windsurf Free | Free | fair use | $7.20 | claude-sonnetgpt-4o | est |
| Windsurf | Windsurf Pro | $15 | fair use | $7.20 | claude-sonnetgpt-4.1gpt-4o | est |
| Claude Code | Claude Code Pro | $20 | 25.0M | $7.20 | claude-sonnetclaude-opus | est |
| Claude Code | Claude Code Max 5× | $100 | 125.0M | $7.20 | claude-sonnetclaude-opus | est |
| Claude Code | Claude Code Max 20× | $200 | 500.0M | $7.20 | claude-sonnetclaude-opus | est |
| Anthropic | Anthropic API | Free | fair use | $12.00 | claude-opusclaude-sonnet | high |
| OpenAI | OpenAI API | Free | fair use | $4.10 | gpt-4.1gpt-4o | high |
Scenario calculator
Adjust your usage pattern to see how much you'd spend on API vs subscription.
Break-even chart
Where API cost crosses each subscription price as hours per day increases.
Dev-Hour Equivalents
Cost per hour of AI-assisted development work, by task type and model.
| Benchmark | Token budget | claude-opus | claude-sonnet | gpt-5.2 | gpt-5 | gpt-4.1 | gpt-4o |
|---|---|---|---|---|---|---|---|
Dev-Hour Equivalent: Light Debug Fix a small bug with minimal context; includes quick diagnosis + patch + unit test. | 25K in 15K out | $0.5775 | $0.3465 | $0.2931 | $0.2093 | $0.1963 | $0.2454 |
Dev-Hour Equivalent: Medium Feature Implement a medium feature across multiple files with validation and tests. | 80K in 60K out | $2.1945 | $1.3167 | $1.1319 | $0.8085 | $0.7392 | $0.9240 |
Dev-Hour Equivalent: Large Refactor Refactor multiple modules with structure changes and regression tests. | 180K in 120K out | $4.5045 | $2.7027 | $2.3042 | $1.6459 | $1.5246 | $1.9058 |
Dev-Hour Equivalent: Long-Context Review Review a large codebase slice (long context) and propose changes with patch notes. | 220K in 70K out | $3.2918 | $1.9750 | $1.5766 | $1.1261 | $1.1550 | $1.4438 |
Use TokenLens in your agent
Install the MCP server so Claude Code, Cursor, or Windsurf can answer pricing questions inline.
// once installed, ask your agent naturally — it calls the right tool and returns a data-backed answer using live TokenLens pricing data.
flikQ/tokenlens-mcp ↗git clone https://github.com/flikQ/tokenlens-mcp.git cd tokenlens-mcp
bun install bun run build
/path/to/tokenlens-mcp with your actual clone pathclaude mcp add tokenlens -- \ env TOKENLENS_DATA_URL=https://tokenlens.flikq.dev/api/data \ node /path/to/tokenlens-mcp/dist/index.js
{
"mcpServers": {
"tokenlens": {
"command": "node",
"args": ["/path/to/tokenlens-mcp/dist/index.js"],
"env": {
"TOKENLENS_DATA_URL": "https://tokenlens.flikq.dev/api/data"
}
}
}
}Use the tokenlens recommend_plan tool. I code 3h/day on claude-sonnet.
Methodology
▸token_cap_estimation
Subscription plans rarely publish exact monthly token caps. We estimate in priority order:
- 1.monthly cap stated directly
- 2.weekly cap × 4.33 (avg weeks/month)
- 3.daily cap × 30.4 (avg days/month)
- 4.multiplier of another plan (e.g. ultra = 20× pro)
- 5.fair use / unknown — shown as —
▸dev_hour_equivalent (DHE)
A DHE represents the token cost of ~1 hour of AI-assisted developer work. Each benchmark profile defines a token budget (input + output) for a typical task type.
adjusted_tokens = budget × verbosity_multiplier × retry_multiplier
dhe_cost = token_cost(adjusted_input, adjusted_output)
▸scenario_modeling
total_tokens = hours/day × days/month × tokens/hour
input_tokens = total × (1 − output_ratio)
output_tokens = total × output_ratio
api_cost = (input / 1M × input_price) + (output / 1M × output_price)
Default output ratio is 0.35. Coding workloads tend higher; review/summarisation lower.
▸confidence_tiers
HIGH — sourced directly from official pricing pages with explicit token counts.MEDIUM — derived from documented limits or community reporting with cross-verification.EST — inferred from marketing language, multipliers, or historical patterns.
▸data_freshness
All data is manually maintained. Each plan entry includes last_verified and effective_from dates. Always verify with the source URL before purchasing.