Release Notes: v0.5.0¶

Release Date: 2026-07-05

Highlights¶

Reasoning Effort Support: Control reasoning depth for o1, o3, and other reasoning-capable models via the unified ReasoningEffort field

New Features¶

Reasoning Effort¶

The OmniLLM provider now supports the ReasoningEffort field for controlling how deeply models reason about problems:

import (
    core "github.com/plexusone/omnillm-core"
    _ "github.com/plexusone/omni-openai/omnillm"
)

effort := core.ReasoningEffortHigh
resp, err := provider.CreateChatCompletion(ctx, &core.ChatCompletionRequest{
    Model:           "o3-mini",
    ReasoningEffort: &effort,
    Messages: []core.Message{
        {Role: core.RoleUser, Content: "Prove that sqrt(2) is irrational"},
    },
})

Supported Values¶

Constant	Value	Use Case
`ReasoningEffortNone`	`"none"`	Disable reasoning
`ReasoningEffortLow`	`"low"`	Quick analysis
`ReasoningEffortMedium`	`"medium"`	Balanced (default)
`ReasoningEffortHigh`	`"high"`	Deep reasoning

Supported Models¶

o1-preview, o1-mini - OpenAI reasoning models
o3-mini - Latest reasoning model
Other reasoning-capable models as released

Dependencies¶

Updated omnillm-core to v0.18.0

Installation¶

go get github.com/plexusone/omni-openai@v0.5.0

Upgrade Notes¶

This is a backwards-compatible release. The new ReasoningEffort field is optional and does not affect existing code.