Skip to main content
Track what’s new in GEPA Omni. For releases across all products, see all changelogs.
​

​
Week of November 2: GEPA Omni 0.3.0 and Fleet Optimize 0.2.0

​
New features

fleet-optimize plugin The marketplace now publishes one plugin, fleet-optimize 0.2.0, with two skills: gepa-omni-skill and the new fleet-optimize-skill. Install it with codex plugin add fleet-optimize@Qredence. See Introduction.Fleet Optimize fleet-optimize-skill launches the latest upstream GEPA optimize_anything with P×N parallel proposals, configurable selection strategies, and grouped batch evaluation. See Fleet Optimize.gepa-omni Python package gepa-omni 0.3.0 exposes optimize, run_optimization, and a typed OmniConfig to applications. Install it from packages/gepa-omni. See API reference.Adaptive auto strategy auto is now the default. It scores the seed, pilots eligible engines, and gives the remaining budget to the best improvement per evaluation, falling back to standalone GEPA for small budgets. The fixed portfolio is opt-in with strategy="omni". See Omni workflow.

​
Updates

  • Upstream GEPA pin. The reflective engine now comes from upstream GEPA pinned to commit 8a2bed96 instead of PyPI gepa==0.1.4.
  • runner_profile replaces agent_backend. The old name stays as an alias for one release. Do not import omni_pipeline or run_omni; use the gepa_omni package. See Engines and backends.
  • Provider key isolation. OPENAI_API_KEY is only sent to api.openai.com; other providers use GEPA_OMNI_API_KEY or GEPA_OMNI_API_KEY_ENV. Endpoints require HTTPS, private addresses need an explicit opt-in, and raw responses are not retained by default. See Providers and endpoint security.
  • New preflight path. Run plugins/fleet-optimize/skills/gepa-omni-skill/scripts/preflight.py. See Quickstart.
​

​
Week of August 3 — GEPA Omni follow-ups

​
Updates

GEPA Omni now targets any OpenAI-compatible Chat Completions endpoint Both the GEPA proposer and the native Omni agent runner (AutoResearch, Meta-Harness, and Best-of-N) now speak a single OpenAI-compatible Chat Completions boundary instead of provider-specific paths, so you can point Omni at any Chat-Completions-compatible model — including the Neon AI Gateway or your own hosted OSS model — without patching the plugin. Fleet Pi compatibility is preserved, token budgets and staged workspace payloads are validated per proposal, and concurrent proposals stay isolated. See the GEPA Omni quickstart and API reference.

​
Bug fixes

  • Interactive setup gate for missing model and base URL. The gepa-omni-skill now prompts once through the skill surface when OMNI_MODEL or OMNI_BASE_URL is missing, keeps the prompted values process-scoped, and never asks for API keys inside chat — keys must still come from the environment or your secret manager. Non-interactive preflight behavior is unchanged, so CI runs still fail fast on missing config. See GEPA Omni quickstart and Gotchas.
​

​
Week of August 3 — Three new products

​
New features

GEPA Omni — Agent Plugins 1.0 optimizer for any scorable text artifact GEPA Omni packages the GEPA Anything optimization stack as an Agent Plugins 1.0 plugin. Write an evaluator that scores a candidate and explains why it failed, and Omni’s two-phase workflow runs three exploration engines (GEPA, AutoResearch, Meta-Harness) in parallel before handing the best candidate to a fresh continuation engine. Install through the Codex marketplace with codex plugin marketplace add Qredence/gepa-omni and invoke the shipped gepa-omni-skill on prompts, programs, configurations, schemas, SQL, regex, plans, or agent instructions. See Introduction to GEPA Omni.
Last modified on November 10, 2026