Browserbase Pricing in 2026: What Are You Really Paying For?

Key Takeaways

• Browserbase uses usage-based pricing across four tiers: Free ($0), Developer ($20 a month), Startup ($99 a month), and a custom Scale plan.
• The monthly fee covers a cloud browser session. Web search, page fetching, agent runs, proxy bandwidth, and model tokens all bill on top of it.
• Agent runs are the hardest line to forecast with Browserbase pricing. Every plan includes a small allotment of three, 15, or 50 runs, then charges pay-as-you-go overage that Browserbase does not price on its public pricing page.
• Browserbase’s free plan has no automatic CAPTCHA solving, caps each session at 15 minutes, and keeps data for seven days, so most production teams upgrade quickly.
• TinyFish is a Browserbase alternative. Search and Fetch are free on every plan, and Agent and Browser run on one credit per agent step or per four minutes of browser time. Agent runs include proxy at no extra charge; browser sessions meter proxy separately by plan.

Browserbase pricing looks simple at first glance: $0 to $99 a month, then usage-based above that.

The headline number covers one thing, a cloud browser session. Everything else a production workflow needs, web search, page fetching, autonomous agent runs, proxy bandwidth, and model tokens, meters on its own line. Two of those lines never appear in the pricing table most teams screenshot.

This guide breaks down every tier, every metered line, the costs that surprise teams in production, real user reviews, and one alternative with a flatter model.

How much does Browserbase cost?

Browserbase runs on usage-based pricing. The public tiers start at $0 and reach $99 a month, then bill overage above the included allotment. Here is the full breakdown, verified against Browserbase's billing documentation.

Two details tend to catch teams off guard. Extract adds $4 per 1,000 calls ($7 with proxies) on top of Fetch. The Model Gateway includes $5 in tokens on the free plan, then bills at market price on every paid tier. Data retention on the Developer plan is seven days, the same as Free, not the 30 days some older comparisons assume.

How TinyFish pricing compares

TinyFish is web-execution infrastructure for AI agents, built on a single credit meter. (Pricing verified July 2026. Credits are the current TinyFish model.) Search and Fetch cost nothing. Agent and Browser draw from the same pool of credits, on either an uncommitted pay-as-you-go plan or a committed subscription.

On cost, three points separate the two platforms. Search and Fetch are free on every TinyFish plan, pay-as-you-go included, where Browserbase meters both after 1,000 free calls each. TinyFish includes model inference, anti-bot, the stealth browser, and Vault and Profiles on every plan. Agent runs include proxy at no extra charge, while browser sessions meter proxy separately by plan. Failed runs and failed fetches cost nothing, since you pay only for successful operations.

One honest trade-off runs the other way. Browserbase's Model Gateway lets a team choose and switch the model that drives its agent logic. TinyFish does not expose model selection; the reasoning is part of the managed platform. Teams that want to bring their own model get that control from Browserbase. Teams that want one flat, all-included price get that from TinyFish.

Start building with TinyFish. Search and Fetch are free on every plan, and signup includes 500 credits with no card required. Get an API key and run a first operation in under a minute.

Every TinyFish plan includes model inference, the stealth browser, anti-bot, agent proxy, Vault and Profiles, and the SDKs, CLI, and MCP (Model Context Protocol) server. Enterprise adds ISO 27001 certification, a research mode, and on-premise deployment.

Browserbase pricing plans: a breakdown

Browserbase pricing is not one number. It is a base fee plus five metered products. Here is what each line actually meters.

Cost line 1: Browser hours and concurrency

Browserbase’s core meter is wall-clock session time, plus how many sessions run in parallel. Concurrency scales from three sessions on Free to 25, 100, and 250 or more across the paid tiers. The free plan includes one browser hour. Developer includes 100 hours, then charges $0.12 an hour. Startup includes 500 hours, then charges $0.10 an hour.

Two ceilings sit underneath the hourly allotment. Session duration is capped on its own: 15 minutes per session on Free, six hours on Developer and Startup. Session creation is rate-limited too, at five a minute on Free, 25 on Developer, and 50 on Startup, so spinning up 100 concurrent sessions on the Startup plan takes about two minutes even though concurrency allows it.

TinyFish meters browser time differently. The Browser API costs one credit per four minutes, with a 60-minute session cap, and concurrency scales at five, 10, and 50 sessions across the pay-as-you-go, Starter, and Pro plans.

Cost line 2: Search API

Browserbase includes 1,000 search calls on every tier, from Free through Startup, then charges $7 per 1,000 calls. The rate limit holds flat at two requests a second across all plans. Worth knowing for anyone comparing search quality: Browserbase's Search runs on Exa, a third-party index, rather than a Browserbase-owned one, which adds a vendor hop to the retrieval path.

TinyFish Search costs zero credits per request. Rate limits scale by plan at 30, 60, and 120 requests a minute, against Browserbase's flat two a second.

On the SimpleQA benchmark TinyFish published, its Search returned both the highest first-result accuracy and the lowest median latency of the five tools tested.

Source: TinyFish Search benchmark. Figures are TinyFish's own published results.

Cost line 3: Fetch API and Extract

Browserbase Fetch meters content-fetch calls. The free allotment is 1,000 calls on Free and Developer, and 10,000 on Startup, then $1 per 1,000 on Developer or $0.50 on Startup, rising to $4 per 1,000 with proxies. Extract adds another $4 to $7 per 1,000. The rate limit is flat at five requests a second.

TinyFish Fetch costs zero credits per URL, handles up to 10 URLs per request, and scales rate limits at 150, 300, and 600 URLs a minute by plan. It returns clean Markdown, HTML, or JSON.

The per-call price is only half the fetch bill. The other half is how many tokens the returned content forces a model to read. In one published TinyFish test across 15 real articles from five publishers, a single Daily Mail article came back from TinyFish Fetch at roughly 4,673 characters, about 92% article content and close to 1,170 tokens with the navigation, widgets, trending links, and ad slots that pad a typical page stripped out. Cleaner extraction means fewer tokens per page, and the model bill is the half of the fetch cost that per-call pricing does not show.

Cost line 4: Agent runs

Agent runs get their own line, and they are the least transparent of the five. Every plan ships a small included allotment: three runs on Free, 15 on Developer, 50 on Startup, and a custom number on Scale. Browserbase treats these as a pay-as-you-go overage line and states plainly that there are "no caps, no cut-offs." Here is the catch. Browserbase publishes an explicit overage rate for Search, Fetch, and Extract, but not for Agents. The Agents row shows only the included counts. A team can run past its allotment; it simply cannot forecast that cost from the pricing page. Browserbase launched Agents in June 2026 as a managed single-call endpoint, and the small allotment reads as a way to sample the product before committing.

TinyFish bills in credits, and the Agent API consumes one credit per step from the same pool as Browser. Concurrency scales at two, 10, and 50 runs by plan, and the per-credit price is published at every tier, so the agent line of a TinyFish bill stays forecastable.

Cost line 5: add-ons

A few costs sit outside Browserbase’s four core meters. Proxy bandwidth is its own line, separate from the proxy surcharge on Fetch: zero gigabytes on Free, one on Developer at $12 per gigabyte overage, and five on Startup at $10.

The Model Gateway includes $5 in tokens on Free, then bills at market price on paid tiers, which is where a team's choice of model shows up on the invoice. CAPTCHA solving is off on Free and automatic on paid tiers, Verified identity is basic on Developer and Startup and full only on Scale, and retention runs seven days on Free and Developer, then 30 days on Startup.

Browserbase hidden costs

The sticker price is the smallest part of a production bill. These are the lines that add up.

• Minimum one-minute session billing. Sessions bill a full minute even when a task finishes in seconds, which inflates cost on short, high-volume jobs.
• Browser-hour overages past the included allotment, at $0.12 an hour on Developer and $0.10 on Startup.
• Proxy bandwidth overages, a separate line from browser hours, at $12 per gigabyte on Developer and $10 on Startup.
• Metered Search and Fetch once the free calls run out: $7 per 1,000 for Search across every tier, up to $4 per 1,000 for Fetch with proxies, and Extract on top.
• Unpriced Agent overage. The included allotment is small, and overage is pay-as-you-go, but Browserbase publishes no per-run rate the way it does for Search and Fetch, so the agent line of the bill is hard to predict.
• Model tokens billed at market price through the Model Gateway beyond the $5 free credit.

One more thing worth knowing before you start: on the free plan, each session caps at 15 minutes, separate from the one-hour monthly allotment. A workflow that needs a single long session, rather than many short ones, hits that ceiling with hours still unused.

Browserbase pros and cons

Browserbase is a mature, well-funded browser infrastructure platform, and it shows in the product. It also spreads cost across enough meters that budgeting takes real work.

Pros

• Stealth and fingerprinting work well out of the box on most targets.
• Session infrastructure is stable and managed, and teams treat it as dependable production infrastructure.
• Observability is a real strength: every run includes a live browser view, session replay, and traces across model and tool calls.
• The Stagehand SDK gives Playwright-level control alongside act, extract, and observe primitives, and plugs into major agent frameworks including LangChain, CrewAI, and Mastra.
• The free tier is enough to prototype.

Cons

• Cost spreads across five metered lines, which makes the monthly bill hard to predict.
• The one-minute billing minimum penalizes short tasks.
• The free tier lacks CAPTCHA solving and Verified identity.
• Browserbase is a build-it-yourself model. You pick the framework (its open-source Stagehand or your own), wire in your models, and run the automation on Browserbase's infrastructure. Teams that want a managed service, one that turns a plain-English goal into a finished result without assembling the stack, will not find that here. Teams that want full control over the framework, models, and logic will count it as a feature.

Stop tracking five meters. TinyFish bills in one credit, with Search and Fetch free, model inference and anti-bot included, and proxy included on agent runs. Compare the plans or start free.

Who is Browserbase best for?

• Teams bringing their own agent framework: Engineers who want raw browser infrastructure and plan to drive it with Stagehand or Playwright get exactly that, plus the option to bring their own model through the Model Gateway.
• Prototypers and early-stage builders: The free tier suits low-volume experimentation before a workflow reaches production scale.
• Stealth-heavy scraping teams: Teams whose main need is fingerprinting and stealth against protected targets are a natural fit for Browserbase's session infrastructure.

Browserbase customer reviews

What users like

Developers consistently praise the stealth layer and the stability of managed sessions. On Reddit, a team running more than 10,000 sessions reported that stealth and fingerprinting held up well across most targets, and a separate thread described the session infrastructure as reliable enough to build on. Reviewers on G2 point to automatic browser scaling as a strength.

What users flag

The most common complaint is the one-minute billing minimum on short tasks, which inflates cost at high volume, and the effort of tracking spend across several metered lines. Some reviewers also note that heavily protected sites remain a challenge.

Alternative to Browserbase: TinyFish

Browserbase gives a team the parts to build a web-automation system. TinyFish is built to be the system. The difference matters most when a workflow stops being simple and starts needing authentication, dynamic pages, and multiple steps that have to hold together.

Goal-based Agent API

With TinyFish, a team passes a URL and a plain-English goal to one of three endpoints: /run for a blocking call, /run-async to start and poll, or /run-sse for a live progress stream. An optional output schema returns typed JSON fields the next step can consume without re-parsing. Browserbase gives you Stagehand, its open-source framework, plus a model of your choosing, and you assemble and run the automation yourself. TinyFish turns one goal into one result on its own managed reasoning.

Authenticated workflows

Browser Context Profiles capture cookies and session storage from a one-time setup session, then replay them on later runs with use_profile: true. Vault, a credential and session capability rather than a standalone product, connects a password manager so an agent can operate inside an authorized session while seeing field labels only, never the credential values themselves, which resolve at the browser layer and are discarded after use. Profiles is currently in beta. One caution worth keeping honest: a run that returns a completed status can still land on a CAPTCHA or access-denied page, so a production workflow should validate the content, not just the status. CAPTCHAs can still be a hard stop.

On cost, the pricing table above covers it: Search and Fetch free, one credit for Agent and Browser, with model inference and anti-bot included. Agent runs include proxy at no extra charge, while browser sessions meter proxy separately by plan.

Test TinyFish on a real workflow. Point it at a live site, pass a plain-English goal, and get structured JSON back. Start with 500 free credits, no card required.

Does Browserbase pricing fit your budget?

Browserbase fits teams that want raw browser infrastructure, will manage five cost lines themselves, and value the option to bring their own model. TinyFish fits teams that want Search and Fetch free and one predictable credit for the rest of the workflow. If your work is mostly stealth-heavy browser sessions you orchestrate yourself, Browserbase earns its place. If it is recurring, multi-step web work where a predictable bill and a single call matter more than model choice, TinyFish is worth a direct test. You can start free with TinyFish, or read the full head-to-head comparison.

FAQs

1. How much does Browserbase cost?

Browserbase costs $0 to $99 a month on its public tiers, then usage-based across five metered lines: browser hours, proxy, Search, Fetch, and Agent runs.

2. Is Browserbase free?

Yes. A free plan includes three concurrent browsers and one browser hour, but no CAPTCHA solving, no Verified identity, and a 15-minute cap per session.

3. How many AI agent runs does Browserbase include?

Three on Free, 15 on Developer, and 50 on Startup, then pay-as-you-go overage, though Browserbase does not publish the per-run rate the way it does for Search and Fetch.

4. Browserbase charge for the Search and Fetch APIs?

Yes. Both include 1,000 calls, then meter on top: $7 per 1,000 for Search, and $0.50 to $4 per 1,000 for Fetch depending on plan and proxy use.

5. What is a cheaper alternative to Browserbase for Search and Fetch?

TinyFish. With TinyFish, Search and Fetch is free on every plan, pay-as-you-go included.