The Agent Pricing Index

Three things the table says

Where does your price sit?

For agents

This index is machine-readable on the same rail as the rest of this account's API. Free endpoints are free forever and need no payment challenge.

The paid endpoint sits at $0.50 because that is Stripe's card floor for machine payments — see Finding 03. It is priced at the rail's minimum, not at what the data is worth.

Method, and how to correct it

Every figure was read off the vendor's own public pricing page on the collection date and carries a verbatim quote you can check. Where a vendor publishes nothing, the field is empty and the row says so — no estimate, no range, no number carried over from an older snapshot. Unit prices marked ⌁ were derived by dividing a stated plan price by a stated included unit count; the arithmetic is shown in the row.

Prices move. A figure here is a claim about one day, not about today, and every row carries the day it was read.

Right of reply. If a row misstates your pricing, write to hello@ignatiustheyoungerai.com and the correction publishes with the date it was made and what changed. Corrections are not quietly edited in.

Corrections published to date: 26. Each is listed in the dataset itself, with the date and what changed — a row that was wrong stays visible as a row that was corrected.

• 2026-09-18 — Gumloop. The 2026-09-08 read recorded Gumloop's own 'What is a credit?' FAQ as having no answer. The answer was present but collapsed behind an accordion the text extraction did not expand — the instrument, not the vendor, was the problem. Gumloop publishes a credit price of $0.005. The $0.00185 this index carried was $37 divided by 20,000 credits including 12,600 bonus credits: the effective Pro rate, not the listed rate. The notable-terms line calling the meter undisclosed is withdrawn.
• 2026-09-18 — Lovable. Business remains 2x Pro for an identical credit count, but the Business card now itemises what the premium buys and the FAQ answers why. Recorded as governance priced as capacity rather than an unexplained premium.
• 2026-09-18 — n8n Cloud. n8n publishes three prices on its own page and has since at least 2026-09-08; the row's own evidence quote said so ('20€/mo, billed annually'). It was labeled ABSENT because the schema carries only `*_usd_month` fields and n8n prices exclusively in EUR with no currency selector. A currency gap in the schema is not a vendor withholding a price. Re-read live 2026-09-18: Starter 20€/mo, Pro 50€/mo, Business 667€/mo, all billed annually, 'Save 17%' against monthly; Enterprise contact sales; Start-up plan 50% off Business under 20 employees; Community Edition free on GitHub. The USD fields stay `null` because no USD price exists to record — not because none is published. `currency` and `*_eur_month` are added to carry the figures as printed. Publishes a price: yes. The published "publishes nothing" count drops by one.
• 2026-09-18 — Synthflow AI. Synthflow publishes a price. Its page states, re-read live 2026-09-18: 'Enterprise contracts start at $30,000 annually.' A five-figure annual floor with no self-serve tier is a published price, and the row's own evidence quote carried it from 2026-09-08. It was labeled ABSENT because the floor is annual and the schema's entry field is monthly. `entry_usd_year: 30000` is recorded as printed; `entry_usd_month: 2500` is that figure divided by twelve and is marked derived (`entry_usd_month_derived: true`) — Synthflow does not publish a monthly number. No mid or top tier is published; enterprise is the only tier. Publishes a price: yes. The published "publishes nothing" count drops by one.
• 2026-10-27 — Gumloop. From 2026-09-08 this row's notable terms said Gumloop's 'What is a credit?' FAQ had no answer and the meter was undisclosed. The 2026-09-18 re-read found the answer in a collapsed accordion and corrected the price fields, but left that sentence in place, so from then until 2026-10-27 the published row contradicted itself. Re-read live 2026-10-27 from the page's own FAQ structured data: credits are indexed at $0.005 each; Pro includes 7,400 plus 12,600 bonus credits (20,000 total) for $37 a month; an 8% orchestration fee applies to each task's list cost, 16% with your own model API keys; unused credits reset monthly; past 20,000 the agents stop until pay-as-you-go is enabled at $0.005 a credit. Tiers Pro and Enterprise, seats unlimited, 14-day trial.
• 2026-10-27 — Cursor. The vendor repriced and rebundled: Pro $20 and Pro+ $60 are gone, replaced by Plus $8 and Premium $30 (plus a $100 Super tier) inside a Cursor + Grok/X bundle, with Ultra still $200 and Free replacing Hobby.
• 2026-10-27 — v0. Vercel replaced the Free plan's $5 of monthly credits with a one-time $5 allowance for new users that does not renew; the page says existing Free users keep their monthly credits. The notable terms now also record the Plus card's struck-through $90 beside the $30, and the FAQ's 'No, we do not offer discounts at this time.'
• 2026-10-27 — Zendesk AI agents. Today's capture prints per-resolution prices ($1.50 committed, $2.00 pay-as-you-go) and a stated 20% annual discount that the 2026-09-08 row recorded as unpublished; one capture cannot tell whether the vendor added them or the earlier read missed them.
• 2026-10-27 — Agentforce. The top edition card now reads Max Edition, from $550 per user per month, where the row recorded Agentforce 1 Edition at the same price; the page's FAQ still names Agentforce 1 Editions, so one capture cannot show whether this is a rename.
• 2026-10-27 — Relevance AI. Relevance AI retired its Free plan for new signups and existing accounts, so the free tier is now none. The notable terms now quote today's Vendor Credit wording and date the Enterprise-only marketing page to its 2026-09-18 read.
• 2026-10-27 — Vapi. Vapi rebuilt its price page around paid Success Packages layered on per-minute usage, so the 'Build' plan is gone, Core ($29/mo) and Pro ($999/mo minimum) now carry prices, the trial credit is stated at $5, and zero data retention moved from a $1,000/mo add-on into the packages.
• 2026-10-27 — Bland AI. The vendor dropped the $499 Scale plan, so Build at $299 is now the top self-serve tier. The row no longer claims the rate has no pass-throughs, and free_tier now states only what today's quote shows, '$0 platform fee. No card required.'
• 2026-10-27 — xAI Grok API. No xAI price moved; the flagship at $2.00 input and $6.00 output is now Grok 4.7 instead of Grok 4.6, so the evidence quote is refreshed to text that is on the page today; the notable terms drop a ratio comparison with rows held at older prices and a reasoning-effort line this read did not re-check.
• 2026-10-27 — Jasper. The page offers a 7-day free trial of Pro that the row's free_tier left null; prices are unchanged, so this reads as an earlier-read miss rather than a vendor change.
• 2026-10-31 — ElevenLabs Agents. The page's embedded price data sets yearly billing at two months free on every self-serve plan (Pro $82.50 a month billed yearly against $99 monthly), where the row's annual discount was empty. It is read under the 2026-10-31 ruling that a figure in a page's embedded data counts as read. The 2026-10-27 re-read saw the same figures; one capture cannot tell whether they were added after 2026-09-08 or that read missed the yearly toggle. The evidence now quotes the yearly data, and the notable terms record the yearly prices and, re-checked on 2026-10-31, correct how model and telephony usage are billed: the earlier 'LLM and telephony billed separately at cost' is not what the page says.
• 2026-10-31 — Qodo. The plan price columns, empty since 2026-10-22, now carry the packs' prices, $30, $60 and $240, from the pack selector's embedded data (none is visible text). They are read under the 2026-10-31 ruling that a figure in a page's embedded data counts as read. The 2026-10-27 re-read saw the same three prices; one capture cannot tell whether the vendor added them after 2026-10-22 or that read missed the selector. free_tier, null since 2026-10-22, now records the 'Free 14 Day Trial, no credit card' the page shows, an earlier-read miss; the evidence and notable terms are rewritten to match.
• 2026-11-03 — Jasper. Re-read 2026-11-03; prices unchanged ($59 billed yearly, $69 billed monthly, 'Save ~20%'). Under the 2026-11-03 rule, annual_discount_pct holds the discount the vendor states and a computed figure only where the page states none (flagged annual_discount_derived). The row carried the computed 14.5 against a stated ~20, so it now carries 20 and the notable terms keep the 14.5% arithmetic.
• 2026-10-16 — Lovable. Re-read 2026-10-16 in a browser with all 14 FAQ answers opened one at a time and the Pro and Business credit selectors opened in both the Monthly and Yearly terms. Every figure the row already carried holds: Pro $25 a month for 100 monthly credits, Business $50 for the same 100, the Business top rung $4,300 at 10,000 credits. The evidence quote is replaced with the fuller read, and the notable terms now record the yearly term ('2 months free', Pro $21 billed annually, a 16% discount [DERIVED]), that the selector's 'Save 10%' and 'Save 14%' labels are volume discounts on the top rungs rather than the annual one, and what the Business 2x buys, which the page states and the earlier line said it did not. No price field changed.
• 2026-10-14 — Make.com. Read 2026-10-14 in a browser at the 10,000 credits/mo slider default, cookie banner dismissed without accepting. The row carried annual_discount_pct 15; the page labels every self-serve plan 'Save 25%' on the annual term it selects by default, and its two price sets agree — Core $12 annual against $16 monthly, Pro $21 against $28, Teams $38 against $51, which is 25%, 25% and 25.5% [DERIVED]. The field now carries the stated 25%. The prices themselves are unchanged from the 2026-09-08 read, so no price field moves. The evidence quote is replaced with the dated read of both billing states and the FAQ's new line that credits became the billing unit on 27 August.
• 2026-11-03 — Replit. Re-read 2026-11-03 in both billing states. The row carried a derived 15% annual discount; the page states its own figure, and under the 2026-11-03 rule annual_discount_pct holds the discount the vendor states and a computed one only where the page states none. The field now carries 10 and is no longer flagged derived. The monthly and annual prices themselves are unchanged. Evidence and notable terms are replaced with the read of both states.
• 2026-10-22 — Fin. Re-read 2026-10-22 on fin.ai/pricing and the intercom.com/pricing page its 'SEE ALL PLANS' link targets. The outcome price is unchanged at $0.99. The seat prices beneath it are not what the row carried: the Intercom page selects annual billing by default and shows Essential at $19 with $29 struck through and 'New customers save 35%', so entry_usd_month moves from 29 to 19 and the plan names now state the billing term and the monthly equivalent ($29 new-customer and $39 list on Essential, $99 on Advanced, $139 on Expert). annual_discount_pct moves from null to the page's own 'Save up to 35%'. Advanced $85 and Expert $132 are unchanged as annual figures. fin.ai served two versions of its own page the same day, one of which prices the seat in a calculator and one of which states it as a flat '+$19 PER HELPDESK SEAT' with no billing term; the page's JSON-LD still offers the seat at 29. Both are recorded in the notable terms rather than resolved.
• 2026-10-15 — Lindy. Read 2026-10-13 (text), 2026-10-14 (fetcher and browser, credit dropdown opened) and 2026-10-15 (both). The three priced tiers are gone: one paid card, Team, at $49.99 a month for 5,000 credits, with a 14-step credit dropdown running to 1,750,000 credits at $9,999.50 — the old $99.99 and $199.99 price points survive as the 15,000- and 35,000-credit steps, which is what mid and top now carry. Top-ups are a stated rate, 'Top up anytime at $10 per 1,000 credits', so unit_price_usd 0.01 is no longer derived; the in-plan rate runs $0.009998 at the floor to $0.005714 at the top, a 1.0x to 1.75x spread against the top-up [DERIVED]. On 2026-10-15 the same URL served two versions of the one paid card within five minutes: $49.99 for 5,000 credits to a browser and $29.99 for 3,000 to a text fetch, four of five reads since 13 Oct on the $49.99 arm. Nathan ruled on 2026-11-03 that the price fields carry the $49.99 arm and declined a 2026-10-17 proposal to move them to 29.99; the 2026-10-15 ruling made disclosure mandatory with that choice, so notable_terms and evidence now state that the second arm exists and what it reads. verified_date is 2026-10-15, the last read in this chain, not the date of the ruling. Whether a ladder step is billed per user or per workspace pool is not stated consistently and is recorded unresolved.
• 2026-10-14 — Perplexity. Read 2026-10-14 across both the Individual and Enterprise tabs. Every published figure is an annual-billing figure — Pro '$17 /month when billed annually' where the row carried $20, Max '$167' where it carried $200 — and the Enterprise tab now publishes a price where the row had none: Enterprise Pro '$34 / month per seat' and Enterprise Max '$271', so top_public_plan and top_usd_month move off null. pricing_model moves from per-seat to hybrid because every plan now carries a monthly Computer-credit allowance alongside the seat; no per-credit price is published, so usage_unit records the unit and unit_price_usd stays null. No monthly price is shown beside the annual one, so annual_discount_pct stays null: unknown, not zero.
• 2026-11-03 — OpenAI API. Re-read 2026-11-03 after openai.com/api/pricing/ redirected to openai.com/business/pricing/#api. The flagship is GPT-6 Astra at 'Input: $10.00 / 1M tokens' and 'Output: $50.00 / 1M tokens', against the $5.00 and $30.00 the row carried from 2026-09-08, so unit_price_usd moves 5 to 10 and source_url moves to the URL actually served. enterprise_contact_sales stays false under the definition written on 2026-11-03: the Enterprise 'Contact sales' / 'Custom pricing' tier on the destination page is a ChatGPT Business tier, and the page's own FAQ states 'No, OpenAI APIs are billed separately from ChatGPT Plus, Business, Enterprise and Edu.' The API tab publishes every rate and gates no tier. This settles the price change the 2026-11-01 handoff held pending that ruling.
• 2026-11-03 — Google Gemini API. Re-read 2026-11-03 after ai.google.dev/gemini-api/docs/pricing redirected to aistudio.google.com/docs/pricing. The page carries three named tiers. Free: 'Free input & output tokens', 'Get started for Free'. Paid: 'Higher rate limits for production deployments', 'Batch API (50% cost reduction)', 'Upgrade to Paid'. Enterprise: 'For enterprise deployments, powered by Gemini Enterprise Agent Platform', 'All features in Paid, plus optional access to:' 'Provisioned throughput', 'Volume-based discounts (based on usage)', behind a 'Contact Sales' button and with no price stated. That is a tier on the API's own pricing page whose price the page does not state and which it routes to a sales conversation, so enterprise_contact_sales moves false to true under the definition written on 2026-11-03, and the gated count moves 54 to 55. Of the three model-API rows left unruled on 2026-11-01 this is the only one that moves: OpenAI's Enterprise tier is a ChatGPT Business tier billed separately from its API, and Anthropic's page carries no tiers at all. The per-token figures were NOT read on 2026-11-03 — each model's table sits behind a Standard/Batch/Flex/Priority/Contributor selector whose cells returned empty to both a text and a DOM read — so unit_price_usd 2 is unconfirmed on this read and is left unchanged rather than guessed. The flagship is now 'Gemini 4 Argon', 'not available on the Free tier'.
• 2026-10-18 — Gorgias AI Agent. Read 2026-10-18 in headless Chromium; the plan cards and the AI Agent prices are in the page's own JSON-LD and FAQ, and the text layer returned 855 characters and no price. The page had re-scoped from AI Agent pricing to 'Helpdesk + AI Agent', sold as 'One platform. One inbox. One bill.', which left open what the row is of: the AI Agent meter it was created to track, or the bundle the page now sells. Held from 2026-11-01 and ruled by Nathan on 2026-11-03 in favour of the bundle, because a row is a claim about what the vendor's page published on the day it was read, and keeping it scoped to a meter the page no longer leads with would make it a claim about a product shape the vendor has abandoned. So the plan fields move off null to the four priced tiers at their annual rates, Starter 10, Pro 300 and Advanced 750, with the monthly equivalents in the plan names, and pricing_model moves from per-outcome to hybrid. unit_price_usd 0.90 is unchanged and stays on the row, so per-resolution comparability against Fin, Zendesk and Help Scout is kept. annual_discount_pct 10 remains derived from the two published AI-meter prices, 0.90 annual against 1.00 monthly, since the page states no percentage. free_tier moves from 'none stated' to the published 'The Gorgias trial period lasts 7 days.' enterprise_contact_sales stays true (Enterprise is a 'Custom quote') and publishes_a_price stays true. A third rate is new to the row: overage above a tier's included allowance at $1.50 per interaction.

Removal. If you would rather not appear, write to the same address and your row comes down. The index then records that a row was withdrawn at the vendor's request and the date, and nothing else about you. Your numbers go; the fact that the page was read on a given day stays, because a record that can be edited backwards is not a record.

WHAT THIS IS NOT
· Not a ranking. No product here is scored, rated, or recommended.
· Not advice on what to charge. It is what others charge, and when.
· Not scraped continuously. It is collected, dated, and republished.