Frontier token prices are 84 percent below march 2023 levels, and open-weight models now carry 56…
frontier token prices are 84 percent below march 2023 levels, and open-weight models now carry 56 percent of gateway token volume, per vercel in september. the pricing war is not about margins, it is about who owns the default.
cheap inference makes the model a commodity and everything around it the product.
Context
Vercel's AI Gateway production index of 17 September 2026 reports that open-weight models ran 56% of AI Gateway tokens in August and 14% of spend, that average price per token fell 23.2% in August, and that it is less than half what it was five months earlier. BenchLM's token price index, updated 1 October 2026, says its index stands at 16 for October 2026, that frontier LLM token prices are 84% below their March 2023 base of 100, with a median blended price of 6.00 dollars per million tokens across 35 flagship models, using blended price of three times input plus output, divided by four.
These are two separate measurements fused in the note. The 56% is Vercel's own gateway traffic for August, not the whole market. The 84% below March 2023 is not in the Vercel post. It comes from BenchLM, a different publisher with a different method, on a page dated after the note, so per Vercel is a misattribution for that figure. Vercel's average price per token can fall because traffic moves to cheaper models, not only because list prices fall. Vercel's post shows Anthropic kept 64% of estimated spend, which cuts against a pure commodity reading. Cheap inference makes the model a commodity is the author's opinion.
The 84 percent here is BenchLM's, and two other notes share that same BenchLM source. Vercel's own figures (a 56 percent open-weight share and a 23.2 percent average per-token fall, traffic-weighted gateway averages) are a separate metric, and the two are kept apart.
Related work
- AI Gateway production index (Vercel) ↗Earlier index, snippet only.
Watch next
- BenchLM's methodology page and the next Vercel monthly index.
Sources
Provenance
The note above is reproduced unedited from the original post, first published on Threads on 18 September 2026 at 17:39 IST. Sources are the papers and datasets the note draws on.
View the original post ↗Embed this note
More notes
The air is now being asked to keep its own ledger
the air is now being asked to keep its own ledger: ecmwf’s aifs compo becomes the first ai model to forecast atmospheric composition globally every three hours, cleanair simulates 365 days of pm2.5 over china in ten seconds, and a unified framework maps six pollutants at one kilometer across the whole country. the air now files its own composition report.
read the note →The current is now being asked to draw its own map
the current is now being asked to draw its own map: china’s langya 2.0 predicts six ocean phenomena including internal waves and mesoscale eddies, a deep net called wenhai resolves eddies globally with air sea flux formulas built in, and scripps infers surface currents from the way temperature patterns deform in satellite images. the ocean now files its own circulation report.
read the note →The soil is now being asked to report its own carbon
the soil is now being asked to report its own carbon: a nix color sensor paired with generative data augmentation predicts soil organic carbon without a lab, random forest drives 74 percent of soil health mapping studies, and sentinel 2 tracks five year carbon change across france and italy from 922 samples. the dirt now files its own carbon account.
read the note →