Product / UX / Orchestration & agents

DSPy

Stanford framework for 'programming - not prompting - language models': compose typed declarative modules (signatures, Predict/ChainOfThought/ReAct) into pipelines, then hand them to optimizers (MIPROv2, BootstrapFewShot, GEPA) that auto-tune prompts and/or weights against a metric.

MIT, no commercial cloud from the project. ~35k stars, ~6.4M PyPI downloads/month.

Openness

5 high confidence

5.0

license: MIT(OSI)
source: public
core-gated: ungated

Fully MIT with no commercial tier.

https://github.com/stanfordnlp/dspy recorded 2026-06-17
MIT, ~35.1k stars

Adoption

4 high confidence

4.0

~6.36M PyPI downloads/month for the dspy package.

https://pypistats.org/packages/dspy recorded 2026-06-17
~6.36M downloads/month

Capability

4 high confidence

4.0

Best-in-class on prompt-optimization/compilation; thinner multi-agent coordination than CrewAI/AutoGen.

https://github.com/stanfordnlp/dspy recorded 2026-06-17
feature set: modules, optimizers, compilation

Unchanged since 2026-07-30 (last edited, not re-checked)