AI tool pricing changes this week: Browser Use gets pricier by default, Factory adds a free plan, Mastra adds a $25 tier
Published · Facts checked against official vendor pages on November 16, 2026 · Rankings are independent of sponsorships
Every Monday we re-read the official pricing page of every tool on ToolTier. This week we checked 146 tools and re-verified 143 of them, including Meta AI and Qwen, which we couldn't confirm before. The big chat and coding plans didn't change. The changes that matter this week are in agent infrastructure, where one default switch can multiply your bill without you touching a setting. Every number below comes from the vendor's own page or docs, read on November 16, 2026.
Tools in this post
The short version
Browser Use: cloud browsers went from $0.02 to $0.06 per browser-hour, and BU 4.0 is now the default agent model at 15 times the token price of bu-fast. If you don't pick a model, you'll pay more.
Factory: Droids now have a $0 Free plan if you bring your own inference (ChatGPT, GitHub Copilot, OpenRouter or your own keys). Paid plans are unchanged.
Mastra: a new $25 Starter plan includes a small dedicated server, the free tier is now called Free, and allowances went up on every plan.
Deals with deadlines: Adobe Firefly Pro and Pro Plus are 50% and 60% off for six months until December 6, and Higgsfield is running a Black Friday sale.
Browser Use: check your default model
Two changes landed at Browser Use this week, and both raise costs. First, the pricing page now lists cloud browsers at $0.06 per browser-hour, billed by the minute. Last Monday it said $0.02. The page's own worked example (100 browser hours, 0.9 GB of home-IP traffic and some BU Fast calls) went from $9.50 to $13.50. Proxy traffic is unchanged at $5/GB through Browser Use's home IPs, or $0.20/GB with your own proxy or none.
Second, browser-use 1.0.6, released on November 16, makes BU 4.0 the default model. If your code calls Agent(task) without passing an llm, it now runs bu-4.0 at $1.80 / $9.00 per 1M input / output tokens instead of bu-fast at $0.12 / $0.60, with high reasoning instead of medium. Browser Use's own release notes say a short test task went from $0.0014 to $0.022. New US Cloud API v4 projects created from November 16 also default to BU 4.0. Older US projects keep GPT-6 Luna, and EU defaults are unchanged.
What to do: if bu-fast was good enough, pass llm='bu-fast' explicitly before you upgrade, or try BU 4.0 Mini ($0.20 / $1.00) for simple tasks. Save BU 4.0 Pro ($3.00 / $15.00) for the hardest flows. Details are on our Browser Use page. If you're building agents that browse, the Agent Builder pack has tool schemas and prompts you can reuse.
Factory: a free plan for Droids, if you bring the model
Factory's pricing page now starts with a Free plan at $0. It works with inference you already pay for: sign in with ChatGPT, GitHub Copilot or OpenRouter, or connect your own API keys or a custom endpoint. Factory's docs say Pro, Plus and Max are the plans where Factory supplies the models, with rolling 5-hour, 7-day and 30-day rate limits.
Paid prices are the same as last week: Pro $20, Plus $100 and Max $200 a month, and Teams $60 a month plus $40 per seat, now listed as up to 10 seats. There's also a new sales-led Business plan with SSO, SCIM, zero data retention and audit logs. If you already pay for ChatGPT or Copilot, you can now try Droids without a second subscription, the way OpenCode, Cline and Amp's free Hobby plan already work. See Factory and our AI coding tools pricing guide.
Mastra: a $25 Starter with a dedicated server
On November 10, Mastra added a $25-a-month Starter plan to its hosted platform. It includes one Small dedicated server (2 GB memory, 1 vCPU, always on), 250K observability events, 10 GB egress and 60-day trace retention. The old $0 tier is now called Free, with 100K events and 15-day retention, and Team stays at $250 a month but now includes two Medium dedicated servers, 10M events and a SOC 2 report.
Serverless agents stay free to run within each plan's allowance. Past it, memory and vCPU cost $0.25 per GB-hour and per vCPU-hour, and observability costs $10 per 100K events. Before, the jump went straight from $0 to $250, so the new tier is good news for small production apps. The open-source framework is unaffected. See Mastra, or compare with LangGraph and CrewAI in our best AI agent frameworks guide.
Discounts running now
These are temporary offers, not list-price changes, so we haven't changed the listed prices:
- Adobe Firefly: Firefly Pro is 50% off (US$9.99 a month instead of US$19.99) and Pro Plus 60% off (US$19.99 instead of US$49.99) for the first six months, if you sign up by December 6. Firefly Premium no longer shows among the individual plan cards, but the page's FAQ still lists it at US$199.99 a month with 50,000 credits.
- Higgsfield: a Black Friday sale of up to 65% off yearly plans. Plus shows $33 a month and Ultra $50 a month billed yearly, against regular monthly prices of $59 and $129.
- Microsoft Copilot: the plans page shows the Individuals plan as "Originally starting from $9.99 now starting from $6.99" a month, with no end date shown.
Newly verified and clarified
Some listings now have prices from an official page for the first time, or clearer details:
- Meta AI is free. Meta's optional Meta One bundles, launched in September, add more Meta AI image and video generation: Core at $7.99 a month and Premium at $19.99 (US prices; availability varies by region).
- Qwen: Qwen Chat is free to use, and the API through Alibaba Cloud Model Studio is pay per token with a free quota for new users.
- Suno: Pro is $10 a month or $8 billed yearly (2,500 credits and 20 song downloads a month), and Premier $30 or $24 yearly (10,000 credits and 60 downloads). The free plan gives 50 credits a day but no downloads.
- Modal now lists pay-per-token Model APIs next to its GPU pricing, for example Kimi K3.1 Pro at $2 / $10 and GLM 5.3 at $1.40 / $4.40 per 1M input / output tokens.
- Amp: students and teachers can get the $20 Megawatt plan for $10 a month.
Deadlines coming up on November 30
Re-checked today on the vendors' docs, these are two weeks away:
- Gemini 4 Argon introductory API price ends. It's $2 / $10 per 1M input / output tokens through November 30, then $4 / $20 from December 1. Our Gemini 4 Argon pricing guide covers the other ways to use it. See Gemini.
- Claude Opus 4.8 fast mode is removed. After November 30, requests to claude-opus-4-8 with speed set to fast return an error. Anthropic points fast-mode users to Claude Opus 5.5. See Claude.
- Claude Sonnet 4.5 retires on the Claude API. claude-sonnet-4-5-20250929 retires November 30, and the recommended replacement is claude-sonnet-5-5. If you pin model IDs in Claude Code or your own agents, check them now.
How we check, and what we couldn't verify
Each Monday we load every tool's official pricing page in a real browser and compare it with last week's copy and with our listing. Where prices hide behind toggles or sliders, we click them or read the page's own data. We only change a price when the vendor's own page shows the change, and tiers never move automatically.
This week we couldn't confirm current prices for 3 tools. Hume AI: its pricing page redirects to the homepage, and its EVI and text-to-speech docs now redirect to the new Audio Decisions API, so we're reviewing that listing. Hailuo AI: the subscribe page still redirects to a homepage that only says "from $9.99/mo". Gamma: its pricing page showed us only British pound prices today. Those listings keep their earlier check dates. The changelog has dated, sourced entries for the Browser Use, Factory and Mastra changes.
FAQ
- Did Browser Use raise its prices?
- Yes. As of November 16, 2026, cloud browsers cost $0.06 per browser-hour, billed by the minute, up from $0.02 on November 9. Proxy traffic is unchanged at $5/GB for home IPs and $0.20/GB without the managed proxy.
- What is the default model in browser-use 1.0.6?
- BU 4.0 (bu-4.0), at $1.80 / $9.00 per 1M input / output tokens, instead of bu-fast at $0.12 / $0.60. Pass llm='bu-fast' to keep the previous model and price.
- Is Factory free now?
- Factory has a $0 Free plan if you bring your own inference: sign in with ChatGPT, GitHub Copilot or OpenRouter, or connect your own API keys. Plans with Factory-supplied models start at $20 a month (Pro).
- How much is Mastra's Starter plan?
- $25 a month since November 10, 2026. It includes one Small dedicated server (2 GB memory, 1 vCPU), 250K observability events, 10 GB egress and 60-day trace retention. Mastra also has a $0 Free plan and a $250 Team plan.
- When does the Adobe Firefly discount end?
- Adobe's US plans page says the offer (50% off Firefly Pro and 60% off Pro Plus for the first six months) ends December 6, 2026.
Free and paid rules for your agents
- AGENTS.md starter for any stack (free)
- CLAUDE.md project memory template (free)
- Agent Builder ($12): 15 production agent designs with tool schemas, guardrails, evals and type-checked code.
- Claude Code Agent Kit ($19): Installable Claude Code plugin: 12 subagents, 8 skills and 3 tested safety hooks.
Related changelog
- 2026-11-16 · Browser Use makes BU 4.0 the default agent model, at 15x the token price of bu-fast
- 2026-11-16 · Browser Use triples its cloud browser price to $0.06 per browser-hour
- 2026-11-16 · Factory adds a $0 Free plan for Droids if you bring your own inference
- 2026-11-10 · Mastra adds a $25 Starter plan with a dedicated server and raises allowances on every plan
Sources
- Browser Use pricing
- GitHub: browser-use 1.0.6 release (Nov 16, 2026)
- Browser Use docs: Models
- Factory pricing
- Factory docs: Individual plans
- Mastra blog: $25 Starter Plan on Mastra Platform (Nov 10, 2026)
- Mastra pricing
- Adobe Firefly plans
- Higgsfield pricing
- Microsoft Copilot plans
- Meta One plans
- Meta: Introducing Meta One (Sept 15, 2026)
- Qwen
- Alibaba Cloud Model Studio: Models
- Suno pricing
- Modal pricing
- Amp pricing
- Gemini API pricing
- Claude docs: Fast mode
- Claude docs: Model deprecations
- Hume API docs
Prices change often. Always confirm on the vendor's site before buying. See the full tier lists and the changelog.