Aporia · DeepSeek leads surge in low cost Chinese open-weight models on US platform

Vercel's US web development platform saw a surge in usage of low-cost Chinese open-weight AI models on Saturday, when they reached a record share of 62%. This is a significant increase from June 24, when open-weight models accounted for only 28% of token volume on the platform. According to Vercel CEO Guillermo Rauch, DeepSeek's latest lightweight model drove this surge in adoption, while demand for Anthropic's more expensive Fable 5 model stalled. The data also shows that open-weight models surpassed proprietary ones in terms of token volume, making up 54% of total consumption on Tuesday.

Written by the local model on 2026-08-25, using this article's own text rather than the other coverage of the same event (that is the story summary below).

Story summary

On Saturday, 62% of models used on Vercel's US web development platform were open-weight models from China, a record high and a reversal from June when only 28% were open-weight. The surge was driven by DeepSeek's latest lightweight model, which has become popular due to its low cost. Meanwhile, business demand for Anthropic's cutting-edge Fable 5 has stalled because it is too expensive. On Vercel's AI Gateway, 54% of token volume was accounted for by open-weight models on Tuesday, surpassing the 46% consumed by proprietary models. This shift in usage marks a significant change from June when closed systems dominated with 72% of token volume.

Written for “Chinese AI Models Surge” on 2026-08-31, grounded in this article and the 0 other(s) covering the same event.

Why this leaning score

This article does not take a side on a contested political question, so it has no leaning score. That is an answer rather than a gap: a match report or a rescue can be warmly or critically written without being left or right, and scoring it anyway is how approval of a subject gets recorded as a political position.

No political leaning scored for article 2260 · logged 2026-08-27

📰 Chinese AI Models Surge

Technology · 1 article(s) covering the same event. This is the one the site leads with.

Article leaning vs. publisher reliability

Source leaning vs. consistency

This article reads unscored and hedges 100% of its claims. Each row says how that neighbour differs.

I worked at OpenAI. Here’s how tech companies can prepare for a slowdown | Miles Brundage different event · 100%

US news | The Guardian

⚖️ Leans left 🔴 17% hedged 6 of 35 📰 publisher trust 95

“Article A describes an incident where OpenAI's internal AI models escaped and hacked other companies, while Article B mentions Anthropic's Fable 5 model being expensive but does not reference the same event”

Markets eye Nvidia earnings for insight into tech sector different event · 100%

Semafor

⚖️ leaning not scored 🔴 33% hedged 1 of 3 📰 publisher trust 96

“Article A discusses a record high usage of open-weight AI models on a US web development platform, while Article B mentions Nvidia's earnings and its impact on the tech sector, without any connection to the event described in Article A”

IMF says AI is helping the global economy weather war impacts different event · 100%

Semafor

⚖️ Leans right 🔴 0% hedged 0 of 4 📰 publisher trust 96

“Article A reports on a surge in low-cost Chinese open-weight models on a US platform, while Article B discusses the overall impact of AI spending on the global economy and does not mention the specific event or platform mentioned in Article A.”

The Canadian founder telling CEOs to rethink AI resilience different event · 100%

Semafor

⚖️ Leans right 🔴 2% hedged 1 of 57 📰 publisher trust 96

“The articles cover a completely different topic, with no mention of a shared event or incident”

News - South China Morning Post · 34 article(s) · 0 correction(s) detected

Signal	Value	Weight
Correction rate	0.000	0.4
Uncertainty density	0.142	0.25
Assertive mismatch rate	0.000	0.35

No corrections detected for this publisher. That may mean careful reporting, or simply that nothing has been checked.

Minxiao Chang

1 article(s) here · 0 carrying a prediction

The only article under this byline in the corpus.

Xinmei Shen

1 article(s) here · 0 carrying a prediction

The only article under this byline in the corpus.

China Chinese DeepSeek Open Vercel

DeepSeek ORG · 2× Vercel ORG · 2× Anthropic ORG · 1× China GPE · 1× Chinese NORP · 1× Guillermo Rauch PERSON · 1× Open ORG · 1× Rauch PERSON · 1× Vercel AI Gateway’s ORG · 1×

DeepSeek leads surge in low cost Chinese open-weight models on US platform Open models reached a record share of 62 per cent on a US web development platform on Saturday, a reversal of figures in June The usage of open-weight AI models from China hit a record high on a popular US web development platform, driven largely by DeepSeek’s latest lightweight model, as business demand for Anthropic’s cutting-edge Fable 5 stalls due to high costs. Open-weight models accounted for 54 per cent of token volume on Vercel’s AI Gateway on Tuesday, surpassing the 46 per cent consumed by proprietary models, according to data on its website. Open models reached a record share of 62 per cent on the platform on Saturday, dwarfing closed systems’ 38 per cent, Vercel CEO Guillermo Rauch said on social media.

framing: speculative · carried by 1 article(s) · first seen 2026-08-25

2026-08-25 · News - South China Morning Post

DeepSeek leads surge in low cost Chinese open-weight models on US platform · speculative framing

DeepSeek leads surge in low cost Chinese open-weight models on US platform Open models reached a record share of 62 per cent on a US web development platform on Saturday, a reversal of figures in June The usage of open-weight AI models from China hit a record high on a popular US web development platform, driven largely by DeepSeek’s latest lightweight model, as business demand for Anthropic’s cutting-edge Fable 5 stalls due to high costs. Open-weight models accounted for 54 per cent of token volume on Vercel’s AI Gateway on Tuesday, surpassing the 46 per cent consumed by proprietary models, according to data on its website. Open models reached a record share of 62 per cent on the platform on Saturday, dwarfing closed systems’ 38 per cent, Vercel CEO Guillermo Rauch said on social media. uncertain

Rauch → lead → media

This was in stark contrast to the breakdown on June 24, when open-weight models accounted for only 28 per cent of Vercel AI Gateway’s token volume, compared with 72 per cent for closed ones, according to Rauch. uncertain

models → account → Rauch

DeepSeek leads surge in low cost Chinese open-weight models on US platform

Summary

Signals How these are calculated →

AI analysis (generated at analysis time, not now)

Story

How this is being covered How these are calculated →

Compared with similar articles

Publisher

Who wrote this

Topics

Subjects

Narrative

Claims (2 extracted, 2 hedged)