OpenAI’s API pricing page lists gpt-rosalind-discovery, a second model in its GPT-Rosalind life-sciences series, at $5.00 per million input tokens, $0.50 for cached input and $25.00 per million output tokens, the same prices it gives gpt-rosalind-research.
The row was pointed out on X by @aitrackerbot at 05:44 UTC on 10 October 2026 and by @testingcatalog at 06:20 UTC. We read the page ourselves at 06:20 UTC on 11 October 2026, and both rows were there.
The page has a section headed “Life sciences models”, introduced with “Our latest Rosalind models.” Its table has two rows, gpt-rosalind-research and gpt-rosalind-discovery, each at $5.00 input, $0.50 cached input and $25.00 output per 1M tokens.
The note under the table names only the research model. It says billing for gpt-rosalind-research began on 5 October 2026 and adds: “Cache-write pricing does not apply to this model.” It limits access to approved internal research through OpenAI’s trusted-access program.
The same note ends: “All eligible organizations will continue to get access to the latest GPT-Rosalind models as they’re released.”
GPT-Rosalind from 11 September to 11 October 2026: OpenAI’s own dated texts, the two X posts that pointed to the gpt-rosalind-discovery pricing row, and our own reading of the pricing page at 06:20 UTC on 11 October.
Business Pill · USAGE-BASED PRICING
A one-minute explainer of usage-based pricing: paying per unit of use, such as per million tokens, rather than per seat. It teaches the general idea only and says nothing about any company in this story.
The key insight: As we read it, OpenAI is turning GPT-Rosalind from one model into a priced product line: a second model sits on the public price list at the same rates as the first, while the texts we read are silent on what it is for.
The two rows of the “Life sciences models” table on OpenAI’s API pricing page, read at 06:20 UTC on 11 October 2026; the note under the table names gpt-rosalind-research only.
What the Two Posts Said
@aitrackerbot wrote that the page “now lists gpt-rosalind-discovery, a GPT-Rosalind model it hasn’t announced yet.” @testingcatalog, whose post had about 38,000 views when read, said the model’s “exact purpose is yet unclear” and asked: “Should it be specifically a drug discovery model?”
That question is @testingcatalog’s. None of the OpenAI texts we read says what gpt-rosalind-discovery is for.
What OpenAI Published the Same Weekend
At 02:00 UTC on 11 October, OpenAI’s developer blog published “Meet Rosalind Workbench”, a research workspace it says is “available in research preview through the ChatGPT app”. It describes guided research tasks across protein design, small-molecule design, genomics and experimental validation, and next-generation sequencing workflows from FASTQ quality control to single-cell analysis.
The post describes two modes, Explore and Research, and says: “Verified organization members can request access on behalf of their organization. Individual access is coming soon.” OpenAI’s help article adds: “You do not need access to the GPT-Rosalind model to use Rosalind Workbench.”
The Workbench post, as we read it, does not name gpt-rosalind-discovery.
Where GPT-Rosalind Stands
OpenAI calls GPT-Rosalind “our frontier reasoning model built to support research across biology, drug discovery, and translational medicine.” An update dated 11 September 2026 on its introduction post says the modelcame out of research preview and that published pricing would take effect on 5 October 2026.
The help article says it is available to eligible organizations with Enterprise or Business agreements through the trusted-access program, in ChatGPT Enterprise, Codex and the API, and that API access is “not available for customer-facing products or external commercial applications at this time.”
The introduction post names Amgen, Moderna, the Allen Institute and Thermo Fisher Scientific among the customers OpenAI is working with. It says GPT-Rosalind outperforms GPT-5.4 on 6 of 11 LABBench2 tasks, and that in a test with Dyno Therapeutics its best-of-ten submissions ranked above the 95th percentile of human experts on a prediction task and around the 84th percentile on sequence generation.
The Structural Read
In the texts we read, the price list moved before the description did. The pricing page carries both rows, but the note under the table, the help article and the Workbench post all describe only GPT-Rosalind in general or gpt-rosalind-research in particular.
The pricing itself is plain per-token billing: identical input, cached-input and output rates for both models, with access tied to OpenAI’s trusted-access program for the research model.
Rosalind Workbench sits on top of the models. OpenAI’s help article says it can be used without access to the GPT-Rosalind model, which separates the workspace from the gated model underneath it.
“All eligible organizations will continue to get access to the latest GPT-Rosalind models as theyโre released.”
Three Implications
LIFE-SCIENCES TEAMS ON THE TRUSTED-ACCESS PROGRAM The pricing page says eligible organizations keep getting the latest GPT-Rosalind models as they are released; it does not say when or how gpt-rosalind-discovery becomes available.
BUDGET OWNERS Both rows are billed per million tokens at $5.00 in and $25.00 out; the note says cache-write pricing does not apply to the research model.
ANYONE READING THE SPECULATION The drug-discovery reading is @testingcatalog’s question, not an OpenAI statement in the texts we read.
The essay argues that what changes between AI paradigms is “where intelligence accumulates and who can capture it”, and describes a Verifier Moat “built by RLVR on domain-specific reward signals”, building “into a new asset class for reasoning-heavy verticals.”
As we read it, a life-sciences model series with its own price list and a gated access program is one way a lab packages a vertical; the texts we read do not say how GPT-Rosalind was trained.
What Is Not Established
OpenAI’s texts that we read on 11 October 2026 do not say what gpt-rosalind-discovery does, who can use it, or whether it can be called yet. The note under the price table names only the research model.
Around 06:25 UTC, OpenAI’s models documentation listed one GPT-Rosalind entry, for gpt-rosalind-research, and a documentation address for gpt-rosalind-discovery returned the general models index. That is the state of the pages when read, not a statement by OpenAI.
Business Engineer Framework
The Four Intelligence Moats
A Business Engineer essay on where intelligence accumulates in the AI stack, and who can capture it.
OpenAI’s pricing page now lists two GPT-Rosalind models at identical prices, $5.00 in and $25.00 out per million tokens, and the second one, gpt-rosalind-discovery, has no description in the OpenAI posts and help pages we read. On the same weekend, OpenAI opened Rosalind Workbench in research preview.
95,000+ executives read Business Engineer for the AI strategy frameworks cited by ChatGPT, Claude, and Perplexity.
A note on sourcing. We read OpenAI’s API pricing page, its 11 October Rosalind Workbench blog post, its GPT-Rosalind help article and its GPT-Rosalind introduction post in full on 11 October 2026, and the two X posts that pointed to the new pricing row. The prices are OpenAI’s; the benchmark results are OpenAI’s own evaluations; OpenAI has not, in the texts we read, described gpt-rosalind-discovery. Nothing here is a forecast, and nothing here is financial or investment advice.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur, HackerNoon and Search Engine People. Find him on LinkedIn and Substack. See how we source.
Scroll to Top
Discover more from FourWeekMBA
Subscribe now to keep reading and get access to the full archive.
