dspy-lite 3.1.3.post2

DSPy with litellm replaced by litelm and numpy made optional

dspy-lite

DSPy with litellm replaced by litelm and numpy made optional.

What this is

• Same DSPy 3.1.3 -- signatures, modules (Predict, CoT, ReAct), optimizers (Bootstrap, MIPRO, SIMBA, GEPA, GRPO)
• litellm (100k+ LOC, 70+ transitive deps) replaced by litelm (2.6k LOC, 2 deps)
• numpy moved from core dep to optional [embeddings] extra
• 88% smaller install (25 MB vs 213 MB, 30 packages vs 72)

Install

pip install dspy-lite

With provider extras:

pip install dspy-lite[anthropic]     # Anthropic
pip install dspy-lite[bedrock]       # AWS Bedrock
pip install dspy-lite[embeddings]    # numpy for embeddings
pip install dspy-lite[all]           # everything

Usage

import dspy

lm = dspy.LM("openai/gpt-4o-mini")
dspy.configure(lm=lm)

predict = dspy.Predict("question -> answer")
result = predict(question="What is DSPy?")
print(result.answer)

Everything from DSPy works -- the import is still import dspy.

What changed from DSPy

	DSPy	dspy-lite
LM routing	litellm (34 MB, 201k LOC)	litelm (87 KB, 2.6k LOC)
numpy	core dep (60 MB)	optional [embeddings]
Install size	213 MB, 72 packages	25 MB, 30 packages
API	unchanged	unchanged

Known limitations

• azure_ad_token_provider passthrough not yet supported (litelm gap)
• Pydantic response_format routing to beta.chat.completions.parse() not yet supported (litelm gap)
• Anthropic citation streaming (provider_specific_fields) not yet supported (litelm gap)

Documentation

Full DSPy documentation: dspy.ai

License

MIT (inherited from DSPy)