Zan Digital › Topics
Pricing Wars
How AI products charge, and what happens to per-seat economics.
- 001Adobe Firefly ARR: Reading the 40% Credit Pack ClaimAdobe says Firefly ending ARR rose 40% in a quarter and AI-first ARR passed $650M. Here is what the filing defines, what it omits, and what to copy.$650M+
- 002Adobe Generative Credits: What $0.005 Buys on Each PlanAdobe generative credits cost about $0.005 each on most plans. We priced each image and video second against API rates, and 1080p Firefly video is the outlier.$0.005
- 003Agent Task Cost: Salesforce, ServiceNow, Microsoft MetersOne 4-step agent task costs $0.40 on Salesforce Flex Credits and $0.136 on Copilot Studio credits. ServiceNow meters it at 25 assists and publishes no price.$0.10
- 004Agentforce Pricing 2026: What the Bundled Credits BuyAgentforce Core at $195 a user bundles 500,000 Flex Credits per org a year, about 25,000 actions. We model all three editions and where each pool runs dry.25,000
- 005Agentforce Pricing: The $2 Conversation Break-Even MathSalesforce charges $2 an Agentforce conversation, or $0.10 a Flex action. The meter beats a $550 seat only below 275 conversations per seat a month.$0.10
- 006Agentforce vs Fin vs Breeze: 3 Prices, 3 MetersFin bills $0.99 per outcome, Breeze $0.50 per resolved chat, Agentforce $0.10 per action. The 10x gap is a unit mismatch. Normalised, the order inverts.10x
- 007AI Agent Costs: Same Task, Up to 30x the TokensRepeat runs of one AI agent task differed by up to 30x in tokens in a 2026 study of 500 coding tasks. Where the variance comes from and how to cap it.Up to 30x
- 008AI Agent Liability: Who Pays When It Gets It WrongStandard AI contracts cap vendor liability at the fees you paid. Deloitte refunded 22% of a A$440,000 report. What buyers can actually recover in 2026.C$650.88
- 009AI Agent Pricing When Nobody Trusts the Agent YetGartner expects over 40% of agentic AI projects to be cancelled by 2027. Trust sets the price, not value. Here is a 4-rung AI agent pricing ladder.46%
- 010AI API Priority Tier Pricing Compared: What 2x BuysOpenAI and Anthropic charge 2x standard for priority tokens and Google 1.8x. None of the 4 per-token tiers states an uptime figure in its own docs.2x
- 011AI Billing: Credits, Tokens and the 20-Week DealCredits, tokens and AI units map to nothing a controller recognises. AI software now takes 16 to 20 weeks to buy, against 7 to 10 weeks for everything else.16-20 weeks
- 012AI Budgeting: Uber's $1,500 Cap and the Exception RuleUber spent its 2026 AI coding budget in about 4 months, then capped tools at $1,500 per engineer a month. A per-seat AI budgeting method with exception rules.$1,500
- 013AI Bundling vs Metering: Both Sides Have CasualtiesMicrosoft bundled Copilot and raised prices up to 45%. Salesforce metered and rewrote pricing 3 times. Both AI packaging strategies have casualties.45%
- 014AI COGS: How to Read a Vendor's Real Gross MarginC3.ai reported 31% GAAP gross margin in FY2026 and 46% non-GAAP. AI COGS hides inference, retrieval, evaluation and human review. How to read the real number.31% vs 46%
- 015AI Contract Negotiation: 6 Clauses That Beat DiscountFederal AI contract terms now specify 72-hour incident notice and 30-day model change notice. Six clauses CFOs are winning in 2026, and what each is worth.72 hours
- 016AI Design Tools Pricing: Five Tasks, Three MetersFigma gives a Professional seat 3,000 AI credits, Adobe Firefly sells 2,000 for $9.99 and Canva counts uses. We price 5 design tasks on all three meters.3,000
- 017AI Price Increases Q4: 7 Pricing Pages, Before and AfterSince July, list prices at 7 AI-heavy vendors moved 0% to 23.6%. The bigger rises hid in limits: Copilot cut included credits 44% with the seat price flat.+16.7%
- 018AI Pricing Changes 2026: 8 Flagged, None Held Up as StatedTrackers logged Runway up 25% and Sybill down 17% this autumn. We checked 8 flagged AI price changes against vendor pages and 0 held up exactly as reported.0 of 8
- 019AI Pricing Changes: 6 Vendors That Deleted Plans (Ledger)Sembly, DocsBot, Z.ai, Zendesk, Microsoft and Rewind removed AI plans instead of repricing them. A dated ledger of 6 deletions, with 9 to 365 days of notice.14 days
- 020AI Provider Contracts Compared: Anthropic, OpenAI, GoogleAnthropic, OpenAI and Google differ on retention, training rights, indemnity and uptime. None of the 3 publishes a credit-backed uptime SLA on standard terms.99.48% v 99.94%
- 021AI Revenue Disclosure: 7 SaaS Metrics, Ranked by RigourWe scored 7 AI revenue metrics from Salesforce, ServiceNow, Adobe, Microsoft and Workday. Microsoft's $37B AI run rate is the biggest number and scores lowest.~$3.9B
- 022AI Shrinkflation: Same Price, Fewer Tokens, Quieter LimitsAI shrinkflation is real but mis-stated. Vendors owe API buyers 60 days notice before a model dies. Subscribers got 31 before limits changed. How to detect it.Under 5%
- 023AI Spend Forecasting Tools: What They Actually Predict79% of enterprises overran AI budgets in the past year. A review of the four AI spend forecasting approaches, what each one sees, and where they break.79%
- 024AI Unit Economics: When Power Users Lose You MoneyAI product gross margins ran 45% in 2025 against the 70-80% SaaS norm. Here is the contribution margin model that shows which customers cost you money.45%
- 025Anthropic Enterprise Pricing: $20 Seats, Uncapped UsageAnthropic cut the Claude Enterprise seat to $20 a month and unbundled every token. Here is the strategic logic, and who absorbs the new volatility.$20
- 026API Price Sunset Calendar: Three Cliffs in 41 DaysThree promotional LLM API rates expire within 41 days, from November 21 to January 1. The dated sunset calendar, and a method to price a workload past each.41 days
- 027B2B Data Providers per Credit: Lusha's 5x Phone RuleLusha charges 1 credit for an email and 5 for a phone, UpLead 1 for both. Contacts run $0.09 to $0.75 on list prices. Cognism and Seamless.AI publish no price.5 credits
- 028B2B Data Providers: $0.37 to $0.75 per Full ContactB2B data providers sell credits, not seats. On list plans an email and phone costs $0.37 at UpLead and $0.52 to $0.75 at Lusha, where a phone is 5 credits.5x
- 029Box AI Units Pricing vs Dropbox Dash: 1,500 Tasks CostedBox Enterprise includes 1,000 AI units a month and Dropbox gives 250 AI credits per user. We cost 1,500 document tasks on each, then on Copilot and Gemini.1,000
- 030ChatGPT and Codex Shared Limits: 3 Plans, No Token CountChatGPT and Codex shared limits arrive in December for 3 plans: Plus, Pro and Business. OpenAI counts usage in tokens but prints no token total for any plan.15-160
- 031ChatGPT Edu Pricing: What 6 University IT Notices RevealChatGPT Edu now bills a seat plus credits worth about $0.04 each. Six university IT notices show the new terms, and what they predict for Enterprise seats.$0.04
- 032ChatGPT Paid Subscribers: The 5% Conversion Rate TimelineChatGPT converts about 5.6% of weekly users to paid: 50M of 900M. Every OpenAI disclosure since 2024 on one timeline, against SaaS freemium benchmarks.5.6%
- 033ChatGPT Pricing 2026: Every Plan Change, Dated and SourcedChatGPT Plus held at $20 all year while Pro 200 drops from 20x to 10x Plus usage on 30 October. Every 2026 ChatGPT plan change, with a date and a source.$20
- 034Claude Pricing 2026: Every Plan and API Rate, CheckedClaude pricing, checked: Pro $20, Max $100 or $200, Team from 2 seats at $20, Sonnet at $2/$10 per million tokens, and why seats now come with API credit.$2/$10
- 035Connecticut AI Law Subscriptions: The Checkout ChecklistConnecticut bars covered AI subscriptions from billing until users accept key terms in writing. It applies above 1 million monthly users. Checklist inside.1 million
- 036Contribution Margin for AI Features: Formula and 3 CasesMicrosoft earned about 44% on each new Intelligent Cloud dollar in FY2026. The contribution margin formula for AI features, tested on three public companies.43.8%
- 037Copilot Business Pricing: 4,000 Credits On by DefaultNew Copilot Business CSP licences bill Copilot Credits by default from 1 December, capped at 4,000 per user. For 50 seats that is up to $2,000 a month extra.$21
- 038Copilot Credits: What Meters, What Stays in the $30 SeatMicrosoft keeps Chat and Office in the $30 Copilot seat and moves Cowork, Code and Autopilot to Copilot Credits with no per-task rates. How to budget anyway.$30
- 039Cost Per Ticket: Human vs AI Agent, With the Full ModelA human ticket costs about $12 on BLS wage data. At a 50% deflection rate, escalation turns a claimed 44% saving into 27%. The full model, with assumptions.$12.04
- 040Cursor Pricing Changes: 6 Plan Shifts Since 2025, DatedCursor pricing moved 6 times since June 2025: 500 requests became $20 of API usage, refunds followed, Auto got metered, and Pro now maps to a $30 tier.500 to $20
- 041Datadog LLM Observability Pricing: $14,125 at 100M SpansDatadog LLM Observability pricing hits $14,125 a month at 100M spans on one stated agent workload. Braintrust Pro costs $3,819. LangSmith prints no rate.$3.50
- 042Datadog LLM Observability Pricing: $14,475 at 100M SpansDatadog LLM Observability costs $14,475 a month at 100M spans in our agent model, against $3,159 on Braintrust. See each assumption and break-even.$14,475
- 043Duolingo Gross Margin 72.9%: AI Calls Under a CentDuolingo gross margin rose to 72.9% in Q3 2026 as Video Call inference fell from $0.30 to under a cent. A 10-Q walk splits model savings from subscription mix.72.9%
- 044Figma Net Dollar Retention: 136% and the AI Credit TestFigma net dollar retention slipped from 139% to 136% in its first full quarter of billed AI credits. We test whether credits or seats drove the expansion.136%
- 045Free Tier Economics: 43% of AI Products Offer Trials43% of AI-native products offer a free trial, against 61% of SaaS, in a 200-product survey. Free tiers are not vanishing. They are being metered instead.43%
- 046Gemini 3.7 Flash Pricing Doubles on January 1, 2027Gemini 3.7 Flash pricing goes from $0.75/$3.75 to $1.50/$7.50 per million tokens on January 1, 2027. Every dated promotional API rate, and how to budget.2x
- 047Gemini Enterprise Pricing: $21, $30 and a Sales CallGemini Enterprise pricing on 29 October 2026: Google lists $21 and $30 starting seat prices, but Plus, Frontline and agent work are priced by sales or by usage.$21
- 048GitHub Copilot Premium Requests: What 20 Engineers PayGitHub Copilot premium requests at $0.04 now survive only on annual plans. On AI Credits, a 20-engineer team at moderate use pays $45.30 a seat on either plan.$0.01
- 049GPT-6 Pricing: Astra, Sol and Luna vs Claude and GeminiGPT-6 Luna lists at $0.10 input and $0.50 output per million tokens, the same as Claude Haiku 5.5. We price 3 workloads and find a 2.5x gap at the top tier.$0.10 / $0.50
- 050GPU Cost Allocation: AWS, Azure and Google Tools ComparedGPU cost allocation from vendor docs: Bedrock bills by IAM role per day, Vertex AI caps each label at 1,000 values, Azure splits by deployment. 7 of 21 native.1,000
- 051Grok Pricing: $8 to $200 in 5 Tiers vs 3 Rival LaddersGrok pricing now runs $8 to $200 across 5 tiers, and Ultra buys 50x Plus usage for 25x the price. A dated tier table against ChatGPT, Claude and Gemini.$8 to $200
- 052HubSpot AI Spend Caps: A 150% Ceiling, Not a ForecastHubSpot's 15 October spend caps set a default credit limit 50% above your monthly credits. That bounds the worst month. It does not predict the next one.50%
- 053HubSpot Breeze Pricing: 3,000 Credits, 60 ResolutionsHubSpot moved Breeze onto credits and outcome pricing in April 2026. At 50 credits per resolved conversation, a Professional plan's 3,000 credits buy 60.50 credits
- 054Hybrid Pricing Hit 37%. The NRR Edge Is 3 PointsHybrid pricing reached 37% of 230 software companies in 2026. The retention edge over subscription pricing is 3 points, and billing overhead is the cost.37%
- 055Intercom Fin Pricing: What Counts as a ResolutionFin bills $0.99 per resolution and passed $100M ARR at 350% growth. Here is how a resolution is defined, where that definition breaks, and what to check.$0.99
- 056Intercom Fin vs Zendesk AI: Cost per ResolutionAt 5,000 tickets a month, HubSpot's Customer Agent costs $2,120, Fin $3,325 and Zendesk AI $4,850 once seats are counted. The resolution rate is half the bill.$0.99
- 057Kustomer Pricing: AI Agent Rates Up 25% After a Blank PageKustomer pricing now lists Concierge at $0.75 per engaged conversation and Envoy at $50 a user, both 25% higher, after a week when the page showed a bare $.+25%
- 058LLM Observability Compared: LangSmith, Braintrust, ArizeLangSmith bills whole traces, Arize bills every span, Braintrust bills gigabytes. One 12-step agent run is 1 billable unit or 12. Compare units, not features.$39
- 059Microsoft 365 Price Increase 2026: Renewal Math by SKUMicrosoft 365 Business Basic rose 16.7% on July 1, 2026, while Business Premium and E1 stayed flat. Per-SKU renewal math and the tier switches that now pay.16.7%
- 060Microsoft Copilot Discount: The 10,000-Seat Price CliffMicrosoft reportedly offers 30% off Copilot above 1,000 seats and 50% above 10,000. At those rates, 9,999 seats cost about $720,000 a year more than 10,000.$15
- 061Net Revenue Retention in 12 Q2 Filings: A 29-Point SpreadNet revenue retention across 12 SaaS Q2 filings runs from 97% at Asana to 126% at Snowflake. Much of that spread is pricing model and definition, not health.126%
- 062Notion AI Credits vs ActiveCampaign: 500 Buys 3 to 500 TasksNotion, ActiveCampaign and Peec each quote about 500 AI credits. At published rates, 500 buys 3 complex Notion tasks, 166 images or 500 email templates.3 to 500
- 063OpenAI Ultrafast Pricing: 6x Rates and the Speed TiersOpenAI Ultrafast bills 6x Standard for up to 300 tokens a second. Every latency price from OpenAI, Anthropic, Google and AWS, and a test for when speed pays.6x
- 064Outcome-Based AI Pricing: 4 of 12 Meters Bill on ResultsWe checked 12 AI agent meters from 7 vendors. Only 4 bill on a publicly defined result, and only Zendesk checks it with more than a customer silence timer.4 of 12
- 065Outcome-Based Pricing Risks: 7 Clauses Buyers Must DemandOutcome pricing shifts risk to the vendor, then lets the vendor keep score. Here are 7 contract clauses that make a disputed AI agent outcome resolvable.24 hours
- 066Per-Seat Pricing Fell to 15%: An Autopsy of the NumberPer-seat pricing did not fall from 21% to 15% in twelve months. That pair of numbers is a Gartner prediction running to 2030. Here is what did move.21% to 15%
- 067Prompt Caching vs Batch Discounts: Effective API RatesClaude Sonnet 5.5, GPT-6.1 Sol and Gemini 4 Argon all charge $0.10 per million cached tokens. Priced on chat, RAG and agent loops, effective rates differ 2.8x.$0.10
- 068Rate Limits as a Pricing Lever: A 5-Test StandardGitHub's premium request multipliers span 173x, from 0.33 to 57. A 5-test transparency standard for AI rate limits, with four vendors graded against it.173x
- 069SaaS Pricing Changes: 1,800 in a Year, What Vendors LearnedThe top 500 B2B and AI companies made 1,800 pricing and packaging changes in 2025, 3.6 each. What actually worked, what backfired, and what to copy.1,800+
- 070SaaS Renewal Increases: Grandfathered Pricing Is EndingAtlassian repriced legacy plans 18% to 40% while list plans rose 15%. Why grandfathered SaaS pricing is ending, and how to negotiate before renewal.18-40%
- 071Snowflake Cortex Pricing vs Databricks: Cost per 1K CallsClaude Sonnet 4.5 summaries cost $9.00 per 1,000 calls on the Anthropic API or Databricks. Snowflake AI_COMPLETE runs $10.80, a 20% markup. The full math.$9.00
- 072State of FinOps 2026: 98% Manage AI Spend, Tools LagState of FinOps 2026: 98% of teams manage AI spend, up from 31% in 2024. We read the sample and map which cost tools see tokens, GPUs and agent runs.98%
- 073Trial to Paid Conversion Rate Benchmarks for AI ProductsA good free-to-paid rate for AI products is 6% to 8%, per a 200-product ChartMogul survey. What ChartMogul, OpenView, Stripe and RevenueCat actually publish.6% to 8%
- 074Unlimited AI Plans: 8 Vendors' Fair Use Clauses Compared4 of 8 AI vendors we checked still sell an unlimited AI plan or feature. None publishes where unlimited stops. Here are the fair use clauses that set the limit.4 of 8
- 075Usage-Based Pricing Problems: The Concessions Buyers Win79% of surveyed enterprises hit AI cost overruns in the past year. Usage-based pricing broke predictability. These are the contract terms buyers now win.79%
- 076Value Metrics That Survive Automation, and How to MigrateSnowflake bills consumption and retains 125% of revenue. Asana bills seats and retains 96%. Here is how to pick a value metric that survives automation.125%