Tamarind Bio pricing, free plan and limits

Free: $0 with 10 jobs a month, the whole model catalog, unlimited storage for job data, email support and runtime caps. Academic: an extended plan with more jobs a month for university, government or nonprofit emails, by application. Premium and Enterprise: no published price, booked through a sales meeting; Enterprise adds custom model hosting, custom security audits, an SLA and SAML SSO. Read on tamarind.bio/pricing in November 2026.

You choose a task such as structure prediction, binder design, antibody design or docking, pick a tool, and submit sequences or structures; Tamarind provisions the GPUs and returns the results in the same job shape for every tool. Pipelines chain tools into one automated run, Batch runs a tool across a list of inputs, and a Model Router recommends a structure-prediction tool from the company's own benchmark evidence. On 11 November 2026 its keyless catalog listed 394 submittable tools, including AlphaFold, RFdiffusion, Boltz, Chai, ProteinMPNN, GROMACS and BindCraft. Its docs name Modal among the service providers that may process customer data.

It is built for computational biologists, protein engineers and structural biology groups who already read scores such as pLDDT and ipTM, working on antibodies and nanobodies, enzymes, peptides or small molecules.

Where an alternative wins: Tamarind runs published models, so a team that wants a platform that learns from its own assay data to design proteins should look at Cradle, which sells through a project form (homepage, read in November 2026) and is not on Hokai yet. Engineers who want to write their own GPU code can use Modal directly, Replicate hosts general models outside biology, and Databricks covers broader data and machine learning work. The wider field sits in the generative AI infrastructure shelf.

More recent additions: a hosted MCP server announced on 27 March 2026, and a wet-lab assay portal and custom tools on 1 May 2026, both from the vendor's changelog. Its blog also describes a protein design agent that follows the binder design workflow Anthropic published after showing Claude design de novo binders, and lists a day-zero release of RFD4-Proteina with the Baker Lab and NVIDIA.

Verdict: pick Tamarind if your lab or biotech needs protein and antibody design tools without running its own GPUs and can start on the Free plan. Look elsewhere if you need prices published beyond that plan, a platform that learns from your own assay data, or data processed outside the United States.