Open weights against proprietary APIs: the cost gap in numbers

Published 2026-08-30 · Prose last reviewed 2026-10-20 · Figures recomputed nightly from the 2026-10-22 catalogue snapshot

A FinOps guide by OptimNow · prices from the live OptimToken catalogue

The argument for open-licensed models arrives as a price argument: publish the weights, let anyone serve them, watch the rate drop. The numbers support that, up to a point, and the point is sharper than the argument suggests. This guide measures the gap three ways: across every model, at matched quality, and at the ceiling where the open models run out.

Openness is not a price tier

Two things about a model are easy to run together. One is what it costs: Premium, Mid-range or Budget, read from the rate card. The other is whose hardware can run it, read from the licence. Open source means OSI-approved terms with no field-of-use limit. Open weights means published weights under a licence that restricts something. Proprietary means API only, and Unknown means no licence is recorded. Of 316 priced models today, 90 are open source, 49 are open weights, 116 are proprietary and 61 carry no established licence. The unknown group stays unknown rather than guessed.

Read straight off, the proprietary median of $6.20 sits 10.3x above the open median of $0.60. That is the number most often quoted, and on its own it proves little. The two groups are not the same models at different prices; they are different models. Proprietary catalogues carry most of the Premium price tier, plus every superseded flagship still on sale behind it. The open groups sit almost entirely in the middle of the market.

The quality-matched comparison, and the ceiling

Arena scores give a common yardstick, on 156 of 316 priced models (85 open, 60 proprietary), scored as of 2026-10-08. Splitting the ranked models into score bands shows why the median comparison misleads. It also shows the finding that matters more than any price ratio on this page.

The best-ranked open model is Kimi K3 at 1488 (Open weights, Kimi K3 License). Above that score sit 8 proprietary models and no open ones: Gemini 4 Argon (1525), Claude Opus 5.5 (1507), Claude Fable 5 (1504), Claude Fable 5.1 (1501), among others. For a workload that needs quality above 1488 on this benchmark, openness is not a lever. No price buys an open model there, because none exists.

Below the ceiling, the gap is large

Inside the overlap the comparison becomes fair, and it is worth running per model rather than per group. For each ranked proprietary model, take every open model with an equal or better Arena score, then the cheapest of those. Of the 60 proprietary models, 52 have such a match. Of those 52 matches, 49 are cheaper, at a median of 27.0x.

One caveat belongs next to that figure. GLM 5.3 Flash is the cheapest qualifying substitute in 36 of the 52 matches, and 4 model makers account for every substitute in the set (Moonshot, Xiaomi, Zhipu, Google). The gap sits in 5 aggressively priced models, not across the open catalogue. Read it as "5 open models are priced far below their measured rank", not as "open licensing is cheaper". It is also fragile: those rates are set by whoever serves the weights, and they move.

Reviewed against the catalogue on 2026-10-20. At publication, on the 2026-08-30 snapshot, the open substitute was cheaper in 35 of 36 matches at a median 17.5x, with one model the cheapest answer in 19 of them. On the 2026-10-22 snapshot it is 49 of 52 at 27.0x, with GLM 5.3 Flash the cheapest answer in 36. The direction held; the median rose, and the paragraph above says why a figure measured on one host's rate does that.

How fragile is measurable. Each price above is the price we quote, which is the rate of one host: a company that runs the model and sells it by the token. Of the open models priced here, 84 are sold by more than one host. Repeat the matched test pricing every open model at its cheapest host and the median gap is 33.5x. Price them at their dearest and it is 11.5x, on the same models and the same day. The proprietary side barely moves under the same treatment: the hosts of a proprietary model charge one rate, a median spread of 1.00x measured below. So treat the 27.0x as the middle of a range you land in by choosing a host, not as a rate card. Which host a price belongs to, and what else the choice changes, is a guide of its own.

What crossing the ceiling costs

Priced on the Knowledge Q&A request the table below costs (2,000 input + 800 output tokens), with Kimi K3 at Moonshot AI's own rate rather than the InferenceNet listing we quote:

At the top of the open range the rule reverses. The best open model, Kimi K3, sits at 1488 and $0.018 a request. Better ranked and cheaper than it are 2 proprietary models: Gemini 3.8 Flash (1497, $0.0045), Gemini 4 Argon (1525, $0.012). On this workload, a buyer picking the strongest open model available pays more for less measured quality than any of them.

What that costs at volume

The Knowledge Q&A preset (2,000 input + 800 output tokens per request), at 1,000,000 requests per month:

Whose rate: Kimi K3 is costed at Moonshot AI's own rate, $3.00 / $15.00; the price we quote for it is InferenceNet's $0.239 / $11.936, a build quantized to fp4. Claude Opus 4.1's quoted rate is Amazon Bedrock's; the model maker publishes no rate of its own. GLM 5.3 Flash's quoted rate is OpenInference's (a build quantized to fp4); Z.AI's own rate is $0.15 / $0.50. Every other row is the model maker's own rate.

The first two rows are the ceiling trade: Kimi K3 is the best open model on offer, Gemini 3.8 Flash the cheapest thing that outranks it. Compare them side by side before the licence decides it. The last two are the widest matched gap in the catalogue, Claude Opus 4.1 against GLM 5.3 Flash at 532x on this workload. Of the 10 widest gaps, 6 sit on proprietary models released at least 12 months before this snapshot. Read a multiple that size as a signal about a model your workloads have outgrown, not as the going rate for proprietary inference.

The licence is not the price, it is the exit

Every price on this page is a hosted API rate, open models included. Publishing weights does not make inference free. It makes the serving market competitive, and someone still runs the GPUs. What the licence buys is optionality: the right to move the same model onto your own hardware if the rate moves, the host withdraws it, or data residency rules out the API.

"Competitive" is measurable. Across the models sold by more than one host, an open model's dearest host charges a median 1.7x its cheapest (n=84). For proprietary models the same figure is 1.00x (n=58): you pay what the model maker charges, wherever you buy it. That is the licence showing up in the price before anyone self-hosts anything. One side of this comparison is a negotiation and the other is a rate card. It also cuts the other way: 80 of the open models priced here have no rate from their own model maker in our data. Every rate quoted for them comes from another host, with no model maker's rate to fall back on if that host withdraws.

That right is not uniform, which is why open source and open weights are kept apart. Of the 85 ranked open models, 55 carry OSI-approved terms (Apache 2.0 or MIT). The other 30 carry a licence that restricts something: an acceptable-use clause, a monthly-active-user threshold, or a non-commercial term. Both are self-hostable. Only one is unconditionally deployable in a commercial product, and the check is a licence read, not a category glance.

Pricing the exit is a separate exercise, and the per-token gap is a poor guide to it. Self-hosting swaps a variable per-token bill for a fixed per-hour one. GPUs bill whether or not they serve traffic, and storage is extra. So is the operations work that TCO calculators leave out: driver and inference-stack upgrades, capacity sizing, preemption handling, model revalidation on every update, on-call. OptimNow's practice reference puts the crossover for a dedicated deployment at roughly 200M to 500M tokens per day, depending on model size and quantisation, with 0.5 to 2 dedicated ML-Ops FTE. Below those volumes, and below FinOps Run maturity, that reference finds the managed API the cheaper answer even when the per-token rate is higher.

What to do with this

• Check the ceiling before the price. If the workload needs quality above 1488, the open question is closed and the decision is which proprietary model.
• Run the substitution per workload, not per catalogue. The median 27.0x is a distribution, not a rate. Your shortlist may sit in the part of it where the gap is small or negative.
• Name the host before you quote the price. An open model's rate is a range across hosts, not a number: the median gap between the cheapest and dearest is 1.7x. A business case built on the cheapest rate is built on one host's pricing decision.
• Read the licence before the roadmap. Open weights and open source are different commercial positions, and the difference only appears in production.
• Treat these as list prices, not spend. Arena is one general benchmark; treat a gap under roughly 20 points as noise.

Figures on this page come from the 2026-10-22 price snapshot and rebuild nightly. Host prices are from the 2026-10-22 read. Filter the pricing table by Openness in the Business view to run this against your own shortlist.

The FinOps reasoning behind this guide comes from OptimNow's open-source practice library: github.com/OptimNow/cloud-finops-skills. If you want the self-hosted question priced against 90 days of your own usage data, OptimNow runs that review.

Data snapshot: 2026-10-22 · Compare all models · Cloud compute pricing · All guides