SAT, SEPTEMBER 26, 2026
Independent · In‑Depth · Practitioner‑Tested
✎ General

China Went Both Ways on Price This Week

DeepSeek reached a $1 billion run rate after raising API prices 4.5x, Alibaba cut voice API prices by up to 95%, and Sanders and Casar introduced a superintelligence ban carrying 20-year prison terms.

By AIToolsRecap September 26, 2026 5 min read 32 views
Home › Articles › General › AI News, 26 Sept 2026: $1B, 95% Off, 20 Years

The day in one line

Two Chinese providers moved in opposite directions on price in the same week, and a US bill proposed prison terms for the thing neither of them is building yet.

DeepSeek: $1 billion run rate, after a 4.5x price rise

The Information reported on 23 September that DeepSeek reached a $1 billion annualised revenue run rate, roughly double its previous figure, citing numbers CEO Liang Wenfeng gave investors. The company is seeking 50 billion yuan - about $7.5 billion - at a 500 billion yuan valuation, about $75 billion.

The sequence is what matters. DeepSeek raised API prices 2.3x to 4.5x in August and quadrupled peak-hour V4 rates. Revenue doubled anyway, because a 4.5x rise on the cheapest inference in the market still leaves it among the cheapest in the market.

If you see a "$500 billion valuation" headline, that is 500 billion yuan with the wrong symbol attached.

Full story and what it means for inference budgets

Alibaba: voice APIs cut by up to 95%

Qwen-Audio 3.1 shipped on 25 September with five models - ASR, TTS and Realtime upgraded, plus new ASR-Next for audio understanding and TTS-Next for audio creation. Cuts run up to 95% on ASR, about 70% on TTS and roughly 85% on realtime. TTS covers 16 languages and 20 Chinese dialect regions; Realtime Plus carries 262K context.

Alibaba published the percentages and not the per-unit prices, and gave no effective date. "Up to 95%" is not a quote.

The five models and whether to switch

Sanders and Casar: 20 years for superintelligence

The Ban Artificial Superintelligence Act, introduced 23 September, creates a cabinet-level Department of Artificial Intelligence, pauses advanced AI development until that department writes safety rules, and carries penalties of up to 20 years imprisonment plus corporate dissolution.

It defines superintelligence as AI that "exceeds human cognitive performance and capabilities across most domains" - prose, with no compute threshold, benchmark or evaluation standard attached. That is the weakness, and it is the one thing the sponsors could fix without changing the bill's intent.

Why the missing threshold is the whole problem

Also worth knowing

  • ChatGPT Voice got plugins. Announced 23 September, rolling out globally from 25 September: email, calendar and Slack plugins work in voice, you can pick between GPT-6 Astra, Sol and Luna, and it runs inside ChatGPT Work on web and mobile.
  • Google Gemini 3.8 added Live Avatar with lip-sync across 97 languages, allowlisted to Gemini Enterprise customers first.
  • OpenEvidence raised $250 million at a $15 billion valuation, up 25% from $12 billion in January.
  • Google Project Suncatcher is scheduled to put four TPUs into orbit on a SpaceX launch on 1 October for a year-long trial.
  • Meta rejected industry-wide AI slowdown proposals, with Zuckerberg declining to join the pacing consensus other labs signed up to this month.

What to watch

Whether DeepSeek raises prices again. A provider that learns a 4.5x increase costs it nothing has been handed a lever, and the next move will tell you whether August was a correction or the first of several.

Tags
AI NewsGenerative AIOpenAIGoogleVoice AI2026
⚑

Spot an inaccuracy?

We verify facts before publishing and correct errors promptly. If something in this article is wrong or outdated, let us know.

Report an error →
💡 AI Tools prompts
Prompt Guide
Best Claude AI Prompts for SEO (2026) — Content, Technical, and Comparison SEO
Claude Sonnet 5 and Opus 5 are strong for SEO work that requires writing quality, structured analysis, and long-form content generation. With 1M context, Claude can analyse an entire site's content structure, compare competing pages, and write complete article drafts in one session. These prompts cover the full SEO workflow: keyword research synthesis, content briefs, on-page optimisation, meta descriptions, technical audit interpretation, and comparison content that ranks above AI Overviews.
Get Prompts →
Prompt Guide
Best ChatGPT Prompts for SEO (2026) — GPT-5.6 and Browse
ChatGPT with GPT-5.6 Sol and Browse enabled is a capable SEO research tool — it can search the live web, analyse SERP results, and synthesise content briefs in a single session. GPT-5.6 Terra at $2.50/M offers a cost-efficient option for high-volume SEO content generation. These prompts are optimised for ChatGPT Plus with Browse, the ChatGPT Work product for larger projects, and the OpenAI API with web_search tool enabled.
Get Prompts →
Prompt Guide
Best Claude Opus 5 and Sonnet 5 Prompts for Writing (2026)
Claude Opus 5 and Sonnet 5 consistently produce the highest-quality long-form writing of any AI model in July 2026 — a lead documented across writing benchmarks and user testing since Claude 3 Opus. With 1M context and 128K output on Opus 5, Claude can write book chapters, complete reports, and long-form content without truncating. Sonnet 5 at $2/$10/M (intro through August 31) is the best value writing model available. These prompts are optimised for claude.ai Pro/Max, Claude Cowork, and the API.
Get Prompts →