MarkTechPost · Oct 2025
Third-party coverage of Anthrogen’s Odyssey launch; explains the 102B-parameter multimodal protein model, the Consensus mechanism, and discrete diffusion.
We're training the next generation of protein foundation models.
Company research
Public company, workplace, funding, and market signals
Updated Jul 29, 2026
Overview
Anthrogen is a San Francisco-based biotechnology/AI research company building Odyssey, a family of multimodal protein foundation models for biologics discovery and development; it combines protein sequence/structure generation, editing, and co-design with an early-access API and collaboration invites.
Primary product
Odyssey, a multimodal protein language model / protein foundation model family
Founded
2024
Headquarters
San Francisco, California, United States
Team size
1-10 employees
Industry
Biotechnology Research
Sub-industry
AI protein foundation models / biologics discovery
Offices
San Francisco, CA
New York, NY
Byron Bay, Australia
Business model
Early-access API
Research collaborations
Biologics discovery platform
Funding
Stage
Seed
Total raised
$4M
Latest round
Seed · Nov 2024
Latest amount
$4M
Nov 2024 · Regen Ventures and BoxGroup
People & investors
Leadership
Co-founder and CEO
Co-founder and CTO
Co-founder and COO
Head of Technical Staff
Founding Member of Platform
Head of Platform
Work & culture
Seed-stage, research-heavy team blending wet-lab and computational work, with a scientist-led culture focused on ambitious protein-design experimentation and rapid model release.
Visa sponsorship
Limited
Compensation
Posted roles advertise roughly $120k-$220k salary plus 0.25%-0.50% equity for Member of Technical Staff and $140k-$220k plus 0.10%-0.50% equity for Robotics Engineer; the jobs pages also mention full health and dental and relocation support when needed, but roles are marked US citizen/visa only.
Benefits
Competitive salary
Equity options
Full health and dental
Relocation costs covered when necessary
Open to new graduates
Product & market
Pricing
Not public; early-access API and collaboration-based access are the only explicit public signals.
Differentiators
- —Odyssey learns jointly from sequence, 3D structure, and functional context
- —Consensus replaces standard attention and is described as more scalable for long sequences
- —Discrete diffusion is used for joint generation and editing
- —The company reports 102B-parameter scale and early-access API availability
- —The team emphasizes large-scale experimental validation and data efficiency
Technology
Python
PyTorch
transformers
discrete diffusion
finite scalar quantizer (FSQ)
consensus mechanism
multimodal protein modeling
Hugging Face checkpoints
Scale & contact
Latest news
Anthrogen · Oct 2025
Official launch page for Odyssey; says the model family supports sequence, structure, and functional-context co-design and offers early-access API access.
bioRxiv · Oct 2025
Research preprint describing Odyssey as a multimodal protein language model family scaled to more than 102B parameters with discrete diffusion and consensus-based architecture.
American Bazaar · Nov 2024
Reports the $4M seed round, names the founders, and describes the company’s rebrand from Arctic Capture to Anthrogen and its carbon-negative chemicals/fuels mission.
Axios · Nov 2024
Reports the seed raise and states the company is using the capital for R&D and dataset building for low-carbon chemicals.
Jobs and careers at Anthrogen
0 jobs at Anthrogen
No open positions
Check back later for new openings