# Chutes AI Pricing: Plans, Limits and Costs (2026) By Dario Espinoza-Blum · 2026-11-07 · Source: https://www.activepieces.com/blog/chutes-ai-pricing-plans-limits-and-costs-2026 --- Chutes AI sells three subscription plans: Base at $3, Plus at $10 and Pro at $20. **Base makes sense for light use** for light, bursty use, because its daily request cap is what you'll actually run into. ## Chutes AI pricing today: plans, prices and limits ![Daily request allowance on each Chutes plan](https://ap-marketing-media.fra1.cdn.digitaloceanspaces.com/uploads/de249a87-86f9-43c8-aca0-d59bee00ee16/chutes-ai-pricing-plans-limits-and-costs-2026-st-356bf236.svg "Source: RPWithAI") Chutes AI pricing is the subscription structure, with Base, Plus and Pro tiers, that sets how many model requests you get per day and what pay-as-you-go rate applies once you exceed that limit. ### What Chutes AI is [Chutes](https://chutes.ai/models) AI is a service that rents you access to AI models over the internet, so you send it a request and it sends back an answer. A "request" is one of those round trips. If you connect it to [Activepieces](https://www.activepieces.com), every step that calls a model counts, so a busy workflow can use up an allowance faster than you expect. The table below lists each plan's monthly price, its daily request allowance and the discount it gives on overage, which is the pay-as-you-go charge once you pass your cap. | Plan | Monthly price | Requests per day | Overage discount | Checked | |---|---|---|---|---| | Base | $3 | 300 | None | 2026-11-07 | | Plus | $10 | 2,000 | 6% | 2026-11-07 | | Pro | $20 | 5,000 | 10% | 2026-11-07 | Ranked by daily allowance, RPWithAI reports: - Base works out to roughly twelve requests an hour around the clock - Plus gives you about 6.7 times Base's requests for 3.3 times the price - Pro gives about 16.7 times the requests for 6.7 times the price ### Chutes AI free tier and pay-as-you-go overage You won't find a free tier in RPWithAI's coverage of the plans, though Chutes publishes per-token rates. Treat both as unknown until you see them on chutes.ai. Past your cap you pay overage, and **Base gets no discount on it**. A Base user who regularly goes over is paying full overage rates on top of the $3, so the plan costs far more than its name suggests. ## How Chutes AI charges beyond the subscription Once your daily request allowance runs out, Chutes AI bills every further request per token, at the rate of the model you called. The $3 stops being your bill and becomes your floor, meaning you cannot budget as if $3 is your maximum cost. On its main TEE models (models that run inside a trusted execution environment), Chutes lists DeepSeek V4 Flash at $0.44 to $1.32 per million tokens, so output tokens cost three times as much as input tokens. ![Chutes per-million-token prices on its main TEE models](https://ap-marketing-media.fra1.cdn.digitaloceanspaces.com/uploads/04e85d69-a11f-45bf-9f79-ced30dbd86e4/chutes-ai-pricing-plans-limits-and-costs-2026-gr-5b9410e6.svg "Source: Chutes") Expect $0.39 to $1.80 on Kimi K2.6, and $0.25 to $2.00 on Qwen3.6-27B, whose wide spread means output-heavy work costs far more than the low figure suggests. Gemma 4 is the cheap option at $0.12 to $0.37, which is the one to check first if you expect to go over. - A daily request allowance that comes with the plan, which is what the $3 actually buys - A pay-as-you-go overage rate that differs for each model - A tier overage discount of 6% on Plus and 10% on Pro, which helps only if you're on those tiers - A separate, lower cached-input rate for repeated prompt text Many small calls burn the allowance faster than a few large ones, because it's counted in requests, not tokens. That suits bursty use, where you hit it hard for an afternoon and then go quiet. It punishes anyone who runs a steady background job. I have no confirmed figure for batch discounts, so don't count on one in your estimate. Price your own workload against the per-model rates above before you subscribe. ## Real monthly costs of the $3 plan For a light user the $3 plan costs about its sticker price, and for a heavy user the bill's set by overage, leaving light users with predictable costs but heavy users facing unpredictable bills. Ask an assistant to explain an error, draft a function or review a change, and you send requests in bursts: a busy afternoon, then nothing for days. That pattern is what the $3 plan suits: - You rarely touch the request limit, so you pay the plan fee and little else - The cost of skipping the plan is paying per-token rates for work that would have fit inside the allowance anyway - A slow month costs you almost nothing extra - A month with a big refactor can push you past the limit, and then overage starts ### Heavy use: a daily batch summarisation job When a job runs every day and summarises a pile of documents, it behaves differently. It sends a steady, predictable stream of requests, and it uses up the allowance early in the month. ![Nearly empty jug beside a calendar with most pages still unturned](https://ap-marketing-media.fra1.cdn.digitaloceanspaces.com/uploads/9782bed1-3f44-40f0-9a79-3677103fde4e/chutes-ai-pricing-plans-limits-and-costs-2026-il-8ea62c23.webp) Everything after that is billed as overage, so the plan fee becomes a small line on a bill dominated by pay-as-you-go charges. A job like this is a repeatable pipeline, and Activepieces can run it: an HTTP step calls the Chutes API for each document, and per-project and per-person caps with usage visibility let you see and limit what the job spends. Activepieces' published pricing page lists the metering as 1 credit per flow run, any number of steps, where Zapier's pricing page bills per task and Make's bills per module. Splitting the job into more steps therefore adds nothing to the automation bill, and the Free plan's 1,000 credits a month covers a once-a-day run. The Chutes overage is still yours to price, as above. ![Activepieces pricing page showing four subscription tiers with features and monthly costs.](https://ap-marketing-media.fra1.cdn.digitaloceanspaces.com/uploads/b5f4b19c-ef62-4a60-90b4-7717913b2053/copy-ai-pricing-2026-plans-costs-and-limits-scre-836b2c9c.webp) ### The break-even request count The chart that follows plots monthly cost against requests per month for the base plan plus overage, the Plus plan and the Pro plan. The break-even is marked where paying as you go becomes cheaper than staying on the base plan. Read it by finding your own monthly request count on the bottom axis. Left of the marked point, the base plan's the cheapest option. Right of it, overage has eaten the saving. ## Value case for the $3 plan Hear the believers out before you drop the subscription. At $3 a month, Chutes AI's plan undercuts the usual paid tiers so far that, for light users, the comparison seems to end there. ### Price per request against the big labs Chutes AI hosts open models, meaning ones whose makers publish them for anyone to run, and sells access to them. Chutes AI's supporters make four arguments. Each rests on a figure: $3 flat, against $8 for [Cursor](https://cursor.com/pricing) Plus (the paid tier of a code-editing assistant) and $20 for Claude Pro (Anthropic's paid chat plan). ![Activepieces pricing page with four subscription tiers showing features and monthly costs](https://ap-marketing-media.fra1.cdn.digitaloceanspaces.com/uploads/5780ea9c-6ecc-45f7-99f2-8707dfb01930/vercel-pricing-in-2026-how-the-costs-scale-scree-22533578.webp) - Cached input at $0.02 to $0.04 per million tokens. A token's roughly a word fragment, and "cached" means text the service has already read once. If your prompts repeat the same long instructions or files, the repeats cost almost nothing. - Many open models behind one plan, so you can switch when one disappoints you instead of paying for a second subscription. - A flat fee that stays $3 however the month goes, so a quiet month costs the same as a busy one. ### Predictable cost and model variety on Chutes AI The second half of the argument is about surprises. A flat price means you can put the cost in a budget before the month starts, and nothing in a normal light month will push the bill higher. Open models also differ in what they do well, so a single plan lets you try several for writing, code or summaries at no extra charge per experiment. ## Why limits and reliability undermine value Only while your usage stays inside the cap does a flat $3 stay flat. Chutes AI has already shown it'll change that cap with little notice, so your costs aren't truly predictable. ### Chutes AI request caps, throttling and retries **300 requests a day** is what the $3 plan buys, according to [RPWithAI](https://rpwithai.com/chutes-new-subscription-plans/). That sounds like plenty until you meet how modern tools behave: a single chat message can trigger several requests behind the scenes, and an automatic retry after a failure counts as another one. ![One letter goes into a slot and many envelopes come out the other side](https://ap-marketing-media.fra1.cdn.digitaloceanspaces.com/uploads/50ffbf6e-19de-40fc-b32c-b6a9672c2775/chutes-ai-pricing-plans-limits-and-costs-2026-il-fac35c86.webp) So 300 requests may be closer to 60 or 100 real exchanges on a busy day. A tool that retries aggressively can burn through the rest before lunch, meaning you lose access by afternoon. ### Chutes AI model availability and plan changes Look at what happened to the free allowance. The chart below shows the entry-level free requests per day at three dates, and the lesson's that **the free floor was removed outright**. Early Access gave 200 free requests a day. The July 2025 $5 deposit kept that same 200, so the paid tier offered no bump in quota. ![Chutes' free entry-level allowance over time](https://ap-marketing-media.fra1.cdn.digitaloceanspaces.com/uploads/7a02e0f0-ad30-4206-95f2-caead8468a19/chutes-ai-pricing-plans-limits-and-costs-2026-st-c4b0f8dc.svg "Source: RPWithAI") On August 4, 2025, the free allowance went to zero, and $3 became the price of 300 requests, so free users faced a sudden paywall. Expect the terms to move again, and keep a fallback provider in mind. ### Chutes AI terms and data handling Before you send anything sensitive, read the data terms. Cheap, shared capacity usually means your prompts pass through hardware you don't control, so the cost of skipping this step is leaking client material or private notes. ![Envelope dropped into a communal mailbox with unseen hands inside](https://ap-marketing-media.fra1.cdn.digitaloceanspaces.com/uploads/6f69f367-f902-4d07-866b-0272fc1c986c/chutes-ai-pricing-plans-limits-and-costs-2026-il-8cd49e73.webp) Check what's logged, who operates the machine running your request, and whether your text can be retained. If you can't find a clear answer, keep confidential work off the plan and use it for throwaway tasks, where a surprise costs you nothing. ## Chutes AI against OpenRouter, Together AI and Groq Once confidential work is off the plan, the next question is whether your throwaway tasks cost less elsewhere. For DeepSeek V3 output they did in our check: **OpenRouter charged $0.89 per million tokens** against $1.20 on Chutes AI, so heavy use of that model cost more on Chutes. ### Chutes AI pricing comparison table For one shared model, DeepSeek V3, the table puts the four providers on the same footing. Read the output price column first, because that's where overage lands. Cells marked "Not confirmed" are figures I couldn't verify, so ask each provider before you commit. | Provider | Entry price | Free tier | Rate limit | DeepSeek V3 output, $/M tokens | Discounts | Check date | |---|---|---|---|---|---|---| | Chutes AI | $3 plan | Not confirmed | Request-based cap | $1.20 | Not confirmed | 2026-11-07 | | OpenRouter | Pay as you go | Not confirmed | Not confirmed | $0.89 | Not confirmed | 2026-11-07 | | Together AI | Pay as you go | Not confirmed | Not confirmed | Not confirmed | Not confirmed | Not checked | | Groq | Pay as you go | Not confirmed | Not confirmed | Not confirmed | Not confirmed | Not checked | The gap between $1.20 and $0.89 only matters once your volume is high enough for overage to show up on a bill. That's the point where the flat price stops protecting you. ### Who should choose Chutes AI or alternatives Pick Chutes AI if your use is occasional and bursty: a few busy days, then nothing. A low fixed fee makes sense there, and you rarely reach the overage rates. Pick OpenRouter if you run steady volume and its per-token rate is lower for your model, as it was for DeepSeek V3 in our check, since the difference compounds with every request. For Together AI and Groq, I won't recommend either on this model without a confirmed price. Get their DeepSeek V3 output rate in writing, then compare it to $0.89 before you sign up. ### Automating usage tracking across providers If you use more than one provider, don't wait for the invoices to find out what you spent. The routine runs in this order: 1. Have your code write one line per request to a shared spreadsheet or log file: date, provider, model, and the token counts each response reports. 2. Once a month, multiply each provider's totals by its output rate. 3. Compare the results. ## What Activepieces does about this The problem in this article is that the $3 sticker price hides what you pay once the 300-request cap is gone, and nobody notices until the invoice arrives. Activepieces is an automation tool, so it can't change Chutes AI's limits. What it can do is put the tracking routine described above into a flow you set up once. A flow calls the Chutes API through an HTTP step, writes one row per request to a sheet (date, model, the token counts the response reports), and a scheduled monthly run multiplies the totals by the per-model rates. You find out whether you've outgrown the Base plan from your own data. The same flow can handle the 402 failure. Add a branch after the call, and if the response is a 402 or any error, the flow can send the request to a second provider, such as OpenRouter at $0.89 per million DeepSeek V3 output tokens against $1.20 on Chutes, or just alert you to top up the balance. That is the fallback this article keeps recommending, built so a changed cap doesn't leave you without a working tool. Both integrations use the generic HTTP step, since I haven't confirmed a dedicated Chutes integration, so check the integrations list before assuming one. The automation bill stays out of the way. Activepieces' pricing page meters 1 credit per flow run regardless of step count, and the Free plan includes 1,000 credits a month, enough for a daily summarisation job with room for the logging run. ![Activepieces pricing page showing four subscription tiers with features and pricing.](https://ap-marketing-media.fra1.cdn.digitaloceanspaces.com/uploads/99e60e08-79ab-4cdf-864e-289013a7463e/neon-pricing-2026-plans-free-tier-limits-and-cos-b089be25.webp) The Community edition is MIT-licensed, so you can self-host it and keep prompts and logs on a machine you control, which matters if you take the data-handling warning above seriously. The Chutes overage is still yours to price, but you will be pricing it from your own logs rather than from a guess. ## Frequently asked questions These four come up right after the plan is set up. Where Chutes AI publishes the answer on its own pages, those pages outrank anything written here, and the checks below take a few minutes. Does Chutes AI have a free tier? Don't assume one. Check the pricing page before you build anything around free access. If you guess wrong, your first sign is a request that fails or a charge you didn't plan for. Providers change free allowances often, so the page you read today is the only version worth trusting. Can I cancel the $3 plan anytime? Cancel from your account's billing settings, and check there whether access continues through the period you already paid for. Do that before you subscribe, not after. A plan you can't easily leave costs more than its sticker price, and a few minutes of reading costs nothing. ### Does Chutes AI charge for failed requests? Before relying on any answer from a third party, read the billing documentation, where Chutes states how errors are counted. Then compare your request log to your first invoice. If a failed call shows up as a charge, you want to learn that in month one, while the amounts are small and a support ticket is easy to file. Which payment methods does Chutes AI accept? The checkout page lists what's accepted for your account and region, and it's the only list that stays current. Look at it before you commit. A payment method that works for the subscription may not be the one you need for pay-as-you-go overage, so confirm both. ## Related reading - [How to Choose a Convex Pricing Plan in 2026](https://www.activepieces.com/blog/how-to-choose-a-convex-pricing-plan-in-2026) - [How to Choose a ClickUp Pricing Plan in 2026](https://www.activepieces.com/blog/how-to-choose-a-clickup-pricing-plan-in-2026) - [WordPress Pricing 2026: Full Plan Comparison](https://www.activepieces.com/blog/wordpress-pricing-2026-full-plan-comparison) ## References - [RPWithAI](https://rpwithai.com/chutes-new-subscription-plans/) - [Chutes](https://chutes.ai/models)