Bend

Health v2 · maintenance only

Adoption not part of the score

23454 stars · 734 forks observed · 2026-10-08

What it is AI-extracted, prompt v1, taxonomy v1, 2026-10-12, confidence 0.97

Bend is a fast, parallel programming language with built‑in dependent types and proof checking. It compiles to native CPU and GPU code and enforces correctness through user‑defined laws.

Use cases

• Write high‑performance parallel programs that run on CPUs and GPUs without writing CUDA kernels.
• Verify AI‑generated code by requiring formal correctness proofs before execution.
• Enforce business or safety rules (laws) in software to prevent AI mistakes.
• Develop applications that need near‑C speed with the ergonomics of a high‑level language.
• Leverage automatic parallelism to scale computations across thousands of cores.

When to choose

• When you need provable correctness for critical or AI‑driven code.
• When maximum performance on both CPU and GPU is required.
• When you want a language that handles parallelism automatically.
• When you prefer a language with dependent types and built‑in theorem proving.

When to avoid

• If you rely heavily on existing libraries that are not available for Bend.
• When the learning curve of dependent types and proof writing is a blocker.
• For small scripts or prototypes where the overhead of proof checking is unnecessary.
• If you need tight integration with ecosystems that only support mainstream languages.

Facets

application · maturity active

programming-language compiler gpu-computing artificial-intelligence machine-learning windows cross-platform cli dependent-types theorem-proving cuda metal ai-safety proofs high-performance proof-checker parallel software-engineering linux macos gpu

Member repositories

For agents

markdown · JSON · MCP: product_card(name="bendlang/bend")