Pricing History Timeline

Stytch removed its free developer tier and set the entry point at $99/mo for up to 1,000 MAU. Startups in early stages were forced to migrate auth systems or pay immediately.

Vercel introduced a new Hobby+ tier between free and Pro. Includes 250GB bandwidth, 3 team members, and commercial use rights. Pro remains at $20/seat.

Anthropic reduced Claude 3.5 Haiku from $1.00/M input to $0.80/M input tokens, and output from $5.00/M to $4.00/M. Matches competitive pressure from GPT-4o-mini and Gemini Flash pricing.

Algolia replaced per-record + per-operation pricing with opaque 'Search Units'. The new model made cost prediction nearly impossible without their calculator. Most users saw 20-40% effective increase.

Render removed the free always-on web service tier. Free static sites and one-time builds remain, but free Node/Docker web services now spin down after 15 min of inactivity (was longer).

GitHub made Copilot available for free to all GitHub accounts: 2,000 code completions/mo and 50 chat messages/mo. No credit card required. Powered by Claude 3.5 Sonnet and GPT-4o.

Netlify moved from fixed-tier pricing to usage-based billing. The Pro plan base dropped but bandwidth, build minutes, and serverless invocations now metered separately. Predictability gone.

Linear increased standard plan pricing after 4 years at $8/user/mo. Added AI features, project views, and SLA tracking as justification.

Oven (the company behind Bun) announced Bun Cloud early access — managed Bun runtime hosting. Pricing not yet public; positioned as Bun-native deploy target competing with Vercel/Cloudflare.

While the price stayed at $20/mo, Cursor reduced the number of 'fast' premium model requests from 500 to 500/mo with stricter counting. Slow requests (queued) became more common during peak hours.

Linear increased per-seat pricing by 25%. Annual plans went from $96 to $120/user/year. Existing annual customers honored until renewal date, then bumped to new pricing.

GitHub announced free Copilot for individuals: 2,000 code completions and 50 chat messages/mo. Available in VS Code, JetBrains, Visual Studio, GitHub.com — no credit card required.

o1 launched at $15/M input + $60/M output (5-6x GPT-4o). o3-mini followed at $1.10/M input + $4.40/M output. New 'reasoning tokens' billed as output tokens.

GitHub introduced a free Copilot tier with 2,000 code completions and 50 chat messages per month. Available to all GitHub users, not just students/OSS maintainers.

Twilio increased outbound SMS pricing by $0.002-0.005/message across US carriers and raised voice per-minute rates. Attributed to carrier fee passthrough but margins also grew.

Notion AI restructured: $10/user as workspace add-on or bundled into new Business plan ($24/user). New Calendar and Mail products bundled, but AI per-seat math got worse for AI-heavy teams.

Convex introduced a Pro tier at $25/mo bridging the gap between free tier and enterprise. Includes 2GB database, 50GB bandwidth, and 10M function calls/mo.

AWS increased Lambda pricing by ~5% and S3 PUT/GET request costs in EU and AP regions. The changes were buried in a blog post with 47 other service updates.

Spaces base subscription stayed at $5/mo (250GB + 1TB egress) but overage pricing increased: storage $0.02/GB (was $0.02), bandwidth $0.01/GB (was $0.01) — minor adjustments and clarifications.

Atlas Serverless rebranded to Flex with new pricing structure. Introduced minimum monthly charges and changed read/write unit pricing. Some workloads saw cost increases of 20-40%.

GitHub launched a free Copilot tier (2000 completions + 50 chats/mo) while quietly raising Business from $19 to $21/user/mo. Enterprise remained at $39/user/mo.

First Cloudflare Pro price increase in 6 years. Business plan stayed at $200/mo. Free plan unchanged. Workers pricing also slightly adjusted.

M0 free shared cluster storage cap stayed at 512MB but connection limit reduced from 500 to 100 in some regions. Free clusters also auto-paused after 60 days inactivity.

Convex Pro changed from flat $25/mo to $25/developer/mo. Production teams with 3+ devs saw bills triple. Free tier remained generous (1M docs, 1 GB).

Docker raised Business plan pricing by ~14%. Docker Team remained at $9/user but added new restrictions on image pulls and build minutes.

Netlify updated bandwidth overage pricing from $20/100GB to $55/100GB on new plans. Existing customers grandfathered temporarily but new signups face higher overage costs.

Sentry Team plan transaction quota reduced from 250K to 100K/mo. Same $26 price. Performance product also re-tiered with new span-based pricing replacing transaction events.

Cursor maintained $20/mo Pro pricing even as underlying model costs dropped 50%+ in 2024. Added Composer, agent mode, and Claude 3.5 Sonnet at no extra cost.

Trigger.dev v3 (complete rewrite) launched with run-based pricing. V2 cloud deprecated. New pricing: free 2k runs/mo, then $10/25k runs. V2 users required to migrate to v3 architecture.

Neon increased free tier from 3GB storage (shared) to 512 MiB dedicated per project with 191.9 compute hours/mo. Also added 10 branches and auto-suspend after 5 min idle.

US carrier surcharges (AT&T, T-Mobile, Verizon) increased throughout 2024. Twilio passes these through, making the effective SMS cost $0.0079 base + $0.005-0.008 carrier fee per message.

GPT-4o mini launched at $0.15/M input and $0.60/M output — cheaper than GPT-3.5 Turbo ($0.50/$1.50) while being significantly smarter. GPT-3.5 Turbo effectively deprecated.

Render introduced bandwidth limits on the free tier (100 GB/mo) and raised the Starter plan. Free static sites remained but free web services got 750 hours/mo cap.

What was bundled APM at $31/host/mo got split into APM ($31), Continuous Profiler ($2/host), Data Streams Monitoring ($25/host). Existing customers grandfathered briefly.

Claude 3.5 Sonnet launched at $3/M input + $15/M output — identical to Claude 3 Sonnet but with significant performance improvements. Effective cost-per-quality dropped 30-50%.

Prisma launched managed connection pooling (Accelerate) and real-time database events (Pulse) as paid products. Free tier includes 6M query requests/mo for Accelerate. Pulse starts at $15/mo.

Salesforce increased Heroku's entry-level pricing. Eco dynos went from $5 to $7/mo and Basic dynos from $7 to $9/mo. Production and Performance dynos also saw 10-15% increases.

Claude 3.5 Sonnet launched at $3/M input and $15/M output — same price as Claude 3 Sonnet but with dramatically better performance, rivaling GPT-4o on most benchmarks.

Stripe restructured Radar pricing: Radar for Fraud Teams went from $0.07/screened to $0.07 + new $0.02 for 3DS authentication. Effective cost increase for merchants using 3D Secure.

MongoDB increased pricing on dedicated clusters (M10+) across most regions by 10-25%. The M10 dev/test cluster went from $57/mo to $65/mo. Shared clusters (M0/M2/M5) stayed the same.

GPT-4o released at $5/M input tokens vs GPT-4 Turbo's $10/M. Performance improved while cost halved. Later reduced further to $2.50/M.

Vercel deprecated standalone Edge Functions and unified all serverless under Functions with optional Edge Runtime. Pricing changed from per-invocation to GB-hours + request count combined.

GPT-4o launched at $5/M input and $15/M output tokens — half the price of GPT-4 Turbo ($10/$30). Also faster and multimodal. Effectively a massive price cut for GPT-4-class performance.

Clerk consolidated plans. Free tier covers 10K MAU. Pro at $25/mo base + $0.02 per MAU above 10K. Removed per-feature pricing in favor of all-inclusive tiers.

Supabase expanded free tier limits: database 500MB→1GB, storage 1GB→5GB, and increased edge function invocations. Pro plan price unchanged at $25.

Vercel clarified that Hobby tier is for personal projects only — any commercial use, ads, or affiliate links forces upgrade to Pro. Tightened bandwidth and serverless quotas at the same time.

Workers Paid ($5/mo) increased CPU time per request from 50ms to 30 seconds (Unbound model unified). Pricing shifted to combined CPU-time + request count rather than separate plans.

Tailscale increased Personal (free) plan from 3 users to 100 users + 100 devices. Premium ($6/user) and Enterprise tiers reserved for SSO, ACLs, audit log integrations.

Neon expanded its free tier significantly — from 1 project to 10 projects, and introduced scale-to-zero by default. Free tier now handles real production workloads for low-traffic apps.

New Team tier fills gap between Pro ($25) and Enterprise. Includes higher connection limits (1000 concurrent realtime), priority support, SOC2 compliance, and daily backups with PITR.

PlanetScale eliminated its free database tier, requiring all users to move to Scaler plan at $39/mo minimum. Thousands of side projects and tutorials suddenly needed migration.

After EU regulatory pressure (Data Act), AWS announced free egress for customers fully migrating off AWS. Doesn't apply to normal egress — only when terminating account.

Vercel's Pro plan shifted middleware execution and serverless compute to a consumption model, removing flat monthly caps. Teams with heavy middleware usage saw 2-5x bill increases overnight.

PlanetScale eliminated the free Hobby tier that offered 5GB storage, 1B row reads, and 10M row writes. Cheapest plan became Scaler at $39/mo. Databases on free tier were scheduled for deletion after a grace period.

Workers free tier increased from 100K requests/day (unchanged) but added 10ms CPU time per invocation (up from 5ms on Bundled). Workers Paid plan stayed at $5/mo for 10M requests.

Redis Ltd changed the license from BSD to dual RSALv2/SSPL, effectively killing the open-source nature of Redis 7.4+. Redis Enterprise pricing remained opaque but trended upward.

Pro plan changed from flat $20/mo for small teams to $20/user/mo. Teams previously paying $20 flat for 3-5 members now face $60-100/mo. Grandfathered pricing expired at renewal.

macOS runner pricing reaffirmed at 10x Linux ($0.08/min on Team). Larger runners (4-core, 8-core) introduced with multiplier pricing. ARM Linux runners launched at premium rates.

Pinecone introduced serverless pricing (per-read/write/storage) replacing flat-fee pod pricing as default. Free tier expanded but production workloads got more variable bills.

Free web services now automatically sleep after 15 minutes of inactivity with 30-60s cold starts. Monthly limit of 750 hours means a single free service can run continuously but multiple cannot.

Vercel restructured Pro plan to per-seat pricing at $20/member/mo and added spend management tools after multiple horror stories of surprise bills going viral on Twitter/X.

Jira, Confluence, and other cloud products saw 5-20% price increases depending on tier. Enterprise customers on annual contracts got hit hardest with multi-year lock-in at new rates.

Unity announced charging $0.20 per install after thresholds. Game developers revolted. Unity partially walked it back but trust was permanently damaged.

Auth0 split free tier into Free (7,500 MAU) and consolidated paid tiers. Critically, SAML/Enterprise SSO moved entirely to Enterprise plan with custom (sales-call) pricing. Custom domains required Essentials+.

Tailscale removed the 20-device limit on the free Personal plan and made it permanently free for individuals. Previously required a paid plan after the trial limit. Teams still need paid plans.

HashiCorp moved Terraform, Vault, Consul from MPL 2.0 to Business Source License. Community forked Terraform as OpenTofu under Linux Foundation.

Replicate moved from charging per inference to per-second GPU runtime. Net effect: cheaper for fast models, more expensive for slow/cold-start workloads. Cold start time now billed.

Following EU pressure and competition from Adyen and Mollie, Stripe reduced European card processing rates from 1.4% + €0.25 to match interchange-plus pricing. Significant for high-volume EU merchants.

Local US number rental went from $1.00 to $1.15/mo. Toll-free went from $2 to $2.15. Combined with 10DLC mandatory registration fees rolled out same year, effective costs climbed 20-30%.

Workers Paid plan changed from $5/mo for 10M requests with $0.50/M overage to $5/mo for 10M requests with $0.30/M overage. Workers KV, R2, and D1 storage pricing also reduced.

Notion rebranded Personal Pro to Plus at $8/mo (down from $10), expanded the free plan to include unlimited blocks for individuals. A rare pro-user move.

Business Starter went from $6 to $7.20/user/mo (20% increase). Standard from $12 to $14.40, and Plus from $18 to $21.60 — all roughly 20% bumps. First major hike since launch.

GitHub (npm's owner) ended the unlimited private package allowance on the free npm tier. Teams relying on private packages for internal tooling now need npm Teams ($7/user/mo) or GitHub Packages.

Docker changed licensing: companies with 250+ employees or $10M+ revenue must pay $5-24/user/mo. Personal and small business use remains free.

Docker announced Free Team orgs would be deleted, including all images. Teams had 30 days to convert to paid ($9/user/mo) or download their images. Initial announcement threatened immediate deletion.

Figma increased Professional plan pricing by 25% and Organization plan from $45 to $75/editor/mo (67% increase). The org tier hike was especially painful for mid-size teams.

After killing free dynos, Heroku launched Eco dynos at $5/mo (1000 hrs across account) and Mini Postgres at $5/mo with 10K rows. Marketed as 'better than free'.

Free workspaces lost access to messages older than 90 days, replacing the previous 10,000 message limit. For small teams this was devastating — years of context gone from search.

R2 went GA in September 2022 with revolutionary pricing: $0.015/GB-mo storage, ZERO egress fees. Instantly the cheapest S3-compatible storage for any read-heavy workload.

Salesforce announced the end of all free Heroku services — free dynos, free Postgres, free Redis — effective November 2022. Millions of hobby projects and student apps lost free hosting overnight.