Thread · Developing story

DeepSeek: pricing and development

The cheapest frontier tier, tracked as it moves — endpoint retirements, benchmark claims, and the shift to charging by the clock.

9 stories · Updated 21 Oct 2026

21 Oct 2026 · Latest DeepSeek Open-Sources DeepInfer, the Engine That Serves Its V4.1 Flash Model on NVIDIA and Huawei Chips

The Apache-licensed code runs one model on NVIDIA SM100 GPUs and Huawei Ascend 950 NPUs, but the cache and load-balancing layers that make it a full service stay private.

17 Oct 2026 DeepSeek Starts Charging Agents for Web Search. Claude Code and Codex Users Pay From 1 November

DeepSeek's new Search and Fetch tools cost US$0.75 per 1,000 requests, and agent tools that use them stop being free on 1 November. An outage the same week cost it a week's refund.

15 Oct 2026 DeepSeek's Funding Round Doubled to 100 Billion Yuan. Its Valuation Did Not Move

Chinese media say Tencent and CATL lead a round that grew from 50 billion yuan in two weeks, priced at the same 500 billion as July, ahead of a planned listing.

15 Sep 2026 DeepSeek Planned to Answer V4 Pro Calls With a Different Model. Then It Backed Down.

The redirect to V4.1 Flash was due at noon on 14 September. DeepSeek now says V4 Pro stays, at the same price. A better model behind the same name can still break things.

11 Sep 2026 DeepSeek Hired Bankers. That Is Not the Same as Filing to List.

Four underwriters for a STAR Market float, a pre-IPO round near $74.5bn, and no prospectus. The reports do not agree on whether the listing lands this year or in 2027.

27 Aug 2026 DeepSeek Is Raising $8bn Two Months After Its First Outside Money

Close to $8bn at around 500bn yuan, up about 11 per cent from June, with a Shanghai listing targeted for 2027. A battery maker is in the round, and the valuation gap with Anthropic...

14 Aug 2026 Everyone Is Cutting Except DeepSeek — and It Now Charges by the Clock

DeepSeek moves V4 to peak and off-peak rates while Google, OpenAI and Anthropic cut. The 1,100 per cent lands on the cheapest token category - and the peak window falls squarely in...

26 Jul 2026 DeepSeek Retired Its Legacy Endpoints — the Migration Is Free, the Headline Price Is Not

Both retired model names resolved to V4-Flash, so a like-for-like migration costs nothing. The number worth checking is the sub-cent cache-hit rate, 50 times below what most worklo...

19 May 2026 DeepSeek V4-Pro Takes the Cost-Performance Crown on SWE-bench at 80.6%, at a Fraction of GPT-5.5's Output Price

DeepSeek's new V4-Pro lands at 80.6% on SWE-bench Verified for $0.87 per million output tokens, a 75% promotional rate, opening a cost-per-fix gap that frontier labs must answer.