Affiliate disclosure: ToolBistro may earn a commission from some links, at no extra cost to you. Facts come from official sources; we do not publish fabricated testing or ratings.

AI Tools

What does LM Studio cost in 2026?

LM Studio pricing is $0, $20, or $100 a month, verified on its pricing page on 13 November 2026. Running models on your own machine stays free at every tier. The two subscriptions buy cloud inference on US-hosted open-weight models with zero data retention, and neither tier publishes how many credits the money includes.

Key facts

What matters

  • Free is a real tier, not a trial: local models through llama.cpp and MLX, the Bionic agent, offline voice transcription and LM Link for five devices.
  • Bionic+ costs $20 a month for US-hosted Kimi K3, GLM 5.3 and DeepSeek V4 Flash with zero data retention and discounted bulk tokens.
  • Pro costs $100 a month and its only published upgrade is 5x usage limits against an allowance LM Studio never states.
  • Ollama sells the same $20 as $60 of monthly credits on a published per-model rate card; LM Studio publishes neither the allowance nor the rates.
  • Local inference is free at both vendors, so the subscriptions buy cloud capacity, not the right to run models on your Mac.

The three LM Studio plans, line by line

LM Studio sells three Bionic plans as of 13 November 2026. The $0 tier is not a trial. It runs local models through llama.cpp and MLX, includes the Bionic agent, offline voice transcription, LM Link pairing for up to five devices, and a limited web search allowance. Nothing there needs a card.

Bionic+ at $20 a month is where cloud inference starts. It adds US-hosted open-weight models (LM Studio names Kimi K3, GLM 5.3 and DeepSeek V4 Flash), zero data retention for those requests, discounted bulk token pricing, and web search plus page extraction.

Pro at $100 a month claims 5x the usage limits of Bionic+, the same zero data retention and bulk discounts, and early access to new features. That is the entire published difference. Teams get the same two tiers through an organization, billed either as pay-as-you-go pooled credits or as seat-based plans.

What Bionic is, and why the plans exist

Bionic is LM Studio's agent app for open models. Inside a project you open separate sessions per task, and each session picks where the model runs: locally on your machine, remotely through LM Link from another device, or in LM Studio's cloud. The cloud option is what the $20 and $100 tiers sell, and LM Studio describes it as Secure Cloud inference with zero data retention.

The split explains the price card. A local session never touches LM Studio's servers, so it stays at $0. A cloud session burns someone else's GPUs, so it gets a subscription. Docs list coding, document work, skills and voice input as the agent's main jobs, which is the same surface Claude Code and Cursor sell, except the models are open-weight.

Is LM Studio still free?

Yes, and more completely than most freemium products: local inference is free at every tier, including the paid ones. LM Studio's billing documentation states that local models and remote LM Link models consume no credits and need no billing set up.

One distinction trips buyers. Bionic is a separate app from LM Studio, per LM Studio's own docs, and the pricing page sells Bionic plans. If the only thing you do is run models on your own machine, the subscription question never comes up.

What you pay for locally is hardware. Model weights have to fit in unified memory or VRAM, so the size of the open model you want sets a one-time hardware bill instead of a monthly one.

The number LM Studio does not publish

Both paid tiers describe usage limits without stating them. Pro's headline upgrade is 5x usage limits, and the base those five times refer to appears nowhere on the pricing page. The credits doc repeats the gap in its own words: Bionic shows usage totals, not a fixed cost per task, and model pricing and request size affect the final spend.

That is an obstacle rather than a nitpick. A buyer choosing between $20 and $100 cannot convert either price into tokens, so the upgrade decision stays guesswork until the first invoice lands. The FAQ only says cloud credits cover usage beyond the allowance included with your plan, and gives no number for that allowance.

Ollama prices the same $20 exactly

Ollama runs the opposite playbook. Its $0 pay-as-you-go tier charges per token at a published rate card. Pro is $20 a month, or $200 a year and $16.67 a month effective, and includes $60 of usage credits each month plus access to larger pro models. Max is $100 with $300 of credits, and Team is $500 with $1,000 of credits shared across unlimited users.

Because the rate card is public, $60 converts into something you can plan around. DeepSeek V4.1 Flash lists at $0.30 per million input tokens and $1.20 per million output, so $60 covers 200 million input tokens or 50 million output tokens. A mixed workload of 10 million input and 1 million output bills $4.20, which works out to about 14 runs on one month of Pro.

The cross-vendor check is possible too. Ollama prices kimi-k3 at $3.00 input and $15.00 output per million tokens, and kimi-k3 is one of the models LM Studio bundles into Bionic+ at an unpublished rate.

Ollama's terms matter as much as its prices. Included credits reset monthly and do not roll over, concurrency is 3 requests on pay-as-you-go and Pro and 10 on Max and Team, and an email fires at 90% of the monthly allowance. Some models carry off-peak rates, with DeepSeek V4.1 Flash falling to $0.15 input and $0.60 output per million.

LM Studio vs Ollama: the price comparison

Neither vendor publishes a comparison against the other, so here is the one buyers keep asking for, built from both pricing pages on the same day.

The short read: LM Studio's $20 is a product subscription with an opaque meter, and Ollama's $20 is a discounted credit pack with a clear one. Both put open-weight models in US data centres and both promise not to train on your prompts. LM Studio states zero data retention for cloud inference by default. Ollama says prompts are never logged or trained on, and that models run primarily in the US with routing to Europe when it needs capacity.

Who should pay for which

  • Running models on your own Mac and nothing else: pay $0 at either vendor. Local inference is unlimited on both.
  • Buying cloud tokens for coding agents and wanting to forecast spend: Ollama, because $60 of credits and public per-model rates let you do the arithmetic before you subscribe.
  • Wanting one desktop agent that switches between local, remote and cloud models: LM Studio Bionic+ at $20, if you accept that its allowance only shows up in the billing dashboard.
  • Running a small team: Ollama Team at $500 a month is explicit about users and credits. LM Studio sends teams to an organization with pooled credits or seat plans and no public price.
  • Handling regulated data: compare clauses, not plans. Both claim zero retention, but only Ollama documents hosting partners and the possibility of European routing.

What moves the bill either way

Token volume decides most of it, then model choice. On Ollama's card GLM 5.3 costs $1.40 input and $4.40 output per million tokens against $0.30 and $1.20 for DeepSeek V4.1 Flash, which is three to four times the price for an identical request count. Cached input is billed separately and far lower: $0.006 per million on Flash and $0.26 on GLM 5.3, which rewards agents that resend the same context instead of restating it.

At LM Studio the visible levers are tier and model class. Pro multiplies an allowance you cannot see, and bulk token discounts apply to volume you cannot price from the page. If the budget matters, ask for the per-token rate before you commit to the $100 tier.

At a glance

LM Studio (Bionic)Ollama
Local models on your own hardwareFree at every tierUnlimited, no subscription
Entry price for cloud models$20/mo (Bionic+)$0 pay-as-you-go, billed per token
Top personal tier$100/mo (Pro, 5x usage limits)$100/mo (Max, $300 of credits)
Included usage creditsNot published$60 on Pro, $300 on Max, $1,000 on Team
Per-model token ratesNot publishedPublished for input, cached input and output per 1M tokens
Concurrent requestsNot published3 on pay-as-you-go and Pro, 10 on Max and Team
Annual billingMonthly only on the pricing page$200/yr on Pro, $16.67/mo effective
Team planOrganization with pooled credits or seats, price not shown$500/mo, unlimited users, $1,000 shared credits
Cloud data handlingUS-hosted, zero data retentionPrimarily US, may route to Europe, no logging or training

FAQ

Is LM Studio free?

Yes. The $0 plan runs models through llama.cpp or MLX, transcribes voice offline, and pairs up to five devices over LM Link. Cloud models need credits, and LM Studio's docs say local and LM Link models need no billing account at all.

How much does LM Studio Pro cost?

$100 a month, against $20 for Bionic+. The published upgrade is 5x the usage limits plus early access to new features. The allowance those limits measure is not published, so neither tier can be costed in tokens from the page.

Is Ollama cheaper than LM Studio for cloud models?

Ollama publishes the arithmetic and LM Studio does not. Ollama's $20 Pro plan includes $60 of monthly credits at a public rate card, with DeepSeek V4.1 Flash at $0.30 per million input tokens. LM Studio's $20 tier lists hosted models and bulk discounts but no allowance and no rate card, so the two can only be compared after a month of real usage.

Related reading

LM Studio vs Ollama compared feature by feature, LM Studio Bionic and its open-model agent, the best local LLM tools, the best VPS for running a local LLM

Sources