Release v0.4.1 · algorithmicsuperintelligence/openevolve

Highlights

Mixed evolution modes (#580, thanks @SUPR1M3): full_rewrite_probability makes a share of diff-based iterations ask for a full rewrite instead. The mode used is recorded on each program and in the evolution trace. Default 0.0.
AlphaEvolve interoperability (#519, #520, #522, #526, #527, #529, #532, #534, #535, #540, #541): opt-in adapters run Google's AlphaEvolve examples and evaluators locally, and export limited settings and Python bundles the other way.
Competitive evolution (#505): alternating competitive evolution with phase checkpoints.
Codex app-server provider (#512): use a ChatGPT subscription as the LLM backend, with no API key.
Per-island setup: per-island system messages (#496) and per-island initial programs (#547).
Runs that generate nothing now fail (#545): if no iteration produces a candidate, OpenEvolve.run raises EvolutionFailedError and the CLI exits with code 1, rather than presenting the seed as the evolved result.
Rewritten README (#580): shorter, and it now leads with research using OpenEvolve. Detailed guides moved to docs/.

--evaluator-args forwards keyword arguments to evaluate() (#506)
logger_prefix and --logger-prefix tell parallel runs apart in the logs (#507)
checkpoint_on_improvement saves a checkpoint when a new best is found (#461)
evolution_trace.include_failures logs failed iterations, including timeouts and crashes, to <trace name>_failures.jsonl (#580)
metadata["llm_model"] records which model generated each program (#580)

Evolution and database

Novelty-rejected programs are dropped from the database (#528)
Island migrants are chosen before any are moved (#524)
New island maps are initialized when resuming checkpoints (#530)
Checkpoint feature statistics are retained during construction (#510)
Inspiration sampling fills the available slots in small islands and bounds neighbors per feature dimension (#508, #509)
Early stopping honors patience with a zero convergence threshold and skips failed evaluations in the plateau count (#566, #502)
Explicit zero iteration budgets are honored (#531)
Very large finite metrics no longer overflow fallback fitness averages (#583)

LLM and workers

Model sampling and request seeds vary per iteration in workers, so runs stay reproducible and parallel iterations don't repeat the same response (#559, #582)
Ensemble weights are validated and safely normalized, zero primary-model weights are preserved, and zero-weight judges are skipped (#570, #533, #567)
CLI backends: Claude Code gets the prompt over stdin and failed responses are rejected, text pipes use UTF-8, and the Codex process stops before its pipes close (#497, #576, #577, #578)
Requests use AsyncOpenAI rather than a sync client in an executor (#513)
Custom clients no longer need a model field (#568)
All top-level config fields reach worker processes (#517), the worker event loop is preserved (#556), and callable evaluators work in subprocess workers (#423)

Parsing and evaluation

Only whole lines count as SEARCH/REPLACE markers (#546)
Full-rewrite code fences handle CRLF, whitespace after the language, literal languages and language-tag fallback (#491, #523, #544)
Java declarations are recognized before generic imports (#560)
Boolean flags survive cascade evaluation, invalid LLM judge metrics are ignored, and partial judge ensembles keep valid feedback (#572, #573, #511)
Embedding vector shapes are preserved after provider failures (#571)
Top-program metrics in prompts use a neutral label (#539)

Evolution trace and visualizer

Children whose parent left the archive are still traced (#581)
Trace payloads are snapshotted, and JSON export metadata can be reused (#579, #575)
HDF5 metadata lists round-trip, and strings that look like JSON stay strings (#580)
The visualizer normalizes checkpoint paths before discovery (#525)

Example evaluators

Evaluators no longer trust values reported by the candidate program (#518, #563)
Non-finite and invalid inputs are rejected across the math and packing evaluators (#551, #552, #553, #555, #564, #565)
The circle packing target is configurable through the environment, and the recorded best result is updated to 2.635977 (#501, #500)

A run where no iteration produces a candidate now raises EvolutionFailedError, and the CLI exits with code 1 (#545).
The invalid llm.model config key is rejected with an actionable error (#503).
Config objects built in code now have their evolution-mode settings validated when passed to OpenEvolve (#580).

Full changelog: v0.4.0...v0.4.1