Enveda raises $311 million to turn nature's chemistry into drugs

Share
Enveda raises $311 million to turn nature's chemistry into drugs

A biotech that reads molecules out of plants and microbes just raised the largest round of the week — and a 2.5B open model showed up to embarrass models twice its size.

Enveda has raised $311 million in a Series E led by Catalio Capital Management, at a $2 billion valuation that doubles what the company was worth twelve months ago. The Boulder, Colorado company does not design molecules from scratch; it searches for them. Its PRISM platform uses machine learning and metabolomics to identify compounds inside plants and microbes, predict their structure and function, and rank them by drug potential — a pipeline its chemists then turn into candidates. Chief executive Viswa Colluru, who founded the company in 2019 after a stint at Recursion Pharmaceuticals, frames the bet as a correction: "We started Enveda on a different premise: the most reliable source of new medicines is the chemistry that living things have been refining for billions of years, and AI finally lets us read it."

The round brings Enveda's total fundraising to more than $845 million, and the money is going into trials rather than more platform work. The company says it has produced 17 candidates to date with three now in human testing. ENV-294, for atopic dermatitis and asthma, is in Phase 2a; ENV-308, aimed at weight-loss maintenance after GLP-1 drugs, cleared a Phase 1 tolerability readout in August; ENV-6946 for inflammatory bowel disease is headed through mid-stage development. BioSpace reports new trials are expected "in the coming months." ICONIQ, T. Rowe Price, Surveyor Capital, Durable Capital Partners, Lightspeed and Lux Capital joined Catalio, alongside a sovereign wealth fund that BioPharma Dive did not name.

The reason this round matters beyond the number is what it says about where AI drug discovery has settled. The 2023–24 pitch was generation — models inventing novel molecules, a framing that ran into a wall of undruggable designs and long preclinical timelines. Enveda's version inverts it: use AI to find what already works biologically, then let chemists finish the job. It is a slower, less cinematic claim, and it is the one a $2 billion valuation just got attached to. We covered the same shift in July, when Basecamp Research raised $140 million to train AI on the planet's DNA — the bet there was on data, here it is on the search itself.


OpenBMB released MiniCPM5-2B, a 2.5-billion-parameter model that beats 4B-class rivals on coding and agent benchmarks — including Alibaba's Qwen3.5-4B. The headline number going around is a 13-point gap, and it is real on exactly one test: SWE-bench Verified, where MiniCPM5-2B scores 46.4 against Qwen3.5-4B's 33.6. It also leads on LiveCodeBench v6 (69.1 to 56.4) and NoLiMa long-context (68.1 to 43.5). But the 13 points is a cherry-pick from a single benchmark. Across OpenBMB's own 34-benchmark average the gap is 2.8 points, 53.9 to 51.1 — and Qwen3.5-4B actually wins on MMLU-Pro (78.0 to 70.8) and MMLU-Redux (88.7 to 84.7).

Independent evaluators are less impressed than the release table. Artificial Analysis scores MiniCPM5-2B at 12 on its Intelligence Index, with Qwen3.5-4B Reasoning at 13 — the composite index does not reproduce the upset, and most of the benchmark numbers are company-reported. What is not in dispute is the packaging: 2.52 billion parameters total, 131,072-token context, Apache 2.0, and traction that speaks for itself at 564,000 downloads in the last month. For a phone-sized model, that is the whole pitch — capability per gigabyte, not per leaderboard.

What to watch: whether Enveda's Phase 2a readouts land before the next funding cycle, and whether anyone runs MiniCPM5-2B against Qwen3.5-4B on a third-party harness rather than the model card.

If a company's own benchmark table is the only place its model wins, should that count as a release or a claim? Write to us via our contact page.

Source: OpenBMB — MiniCPM5-2B model card

Get SiliconNoon in your inbox

One email a day with the AI stories that matter — what happened, the numbers, and why it matters. No spam, unsubscribe in one click.

Read more

Altis Labs raises $25M to make an AI scan read a trial endpoint

Altis Labs raises $25M to make an AI scan read a trial endpoint

A cancer-trial startup wants its model's reading of a CT scan to be the number a drug is judged by — and Apple's Siri home hub is already in employees' hands. Altis Labs raised $25 million to make an AI reading of a CT scan the thing a cancer trial is measured against. The round, announced September 17 and co-led by OrbiMed and Qiming Venture Partners USA, funds a company whose product is not a drug and not a chatbot: IPRO, a model that reads routine CT scans taken during oncology trials and…

Tandem Health raises $100M for Europe's AI clinic operating system

Tandem Health raises $100M for Europe's AI clinic operating system

Monday afternoon's money went into the least glamorous part of the AI stack — the part with a regulator attached. Stockholm's Tandem Health landed Europe's biggest clinical-AI round to date, Waymo opened its doors in Las Vegas, and Akuity made the case that agents need an identity before they get a production deploy. Tandem Health raised a $100M Series B led by the Scaleup Europe Fund, the €5 billion vehicle the European Commission put EQT in charge of this year.

More molecules, same failures: the crack in AI drug design

More molecules, same failures: the crack in AI drug design

China's AI-for-drug-discovery crowd spent Sunday at the Pujiang Innovation Forum arguing about the thing benchmarks don't measure — what happens after the model proposes. And the country's biggest banks are quietly building a lending product around the same question: is AI usage real enough to lend against? "The more you generate, the more you fail." That was Ding Sheng, founding dean of Tsinghua's school of pharmaceutical sciences, at the forum's Future Medicine track on September 13, and the…

Amodei: AI will cure most human diseases within a decade

Amodei: AI will cure most human diseases within a decade

Two stories out of the labs' pre-IPO week: Anthropic's CEO staked public trust on a medical moonshot, and OpenAI kept losing the engineers who built it. Dario Amodei published his most personal case for Anthropic yet — and set a deadline: AI will cure most human diseases within five to ten years, with early results promised "in the coming months." In a rare long-form post, Anthropic's CEO answered Silicon Valley's loudest critic, investor Gavin Baker, who had said (citing "multiple credible…

Qualcomm buys the MoveIt robotics stack and promises to keep it open

Qualcomm buys the MoveIt robotics stack and promises to keep it open

The chip vendor bought the middleware that most robots in research and industry are actually planned with — and the promise to keep it open is now the part that matters. Meta shipped the smallest AI hardware at Connect, and OpenAI took its ad business into Southeast Asia. Qualcomm will acquire PickNik Inc., the Boulder company that has stewarded the MoveIt manipulation framework since its Willow Garage days, and says MoveIt 1 and MoveIt 2 will remain open source under their existing license.