Update OpenAI models to use direct API instead of Zen routing

Agent-Logs-Url: https://github.com/thisguymartin/devkit/sessions/3b80a853-067a-4f41-b26c-58f4a498b4dc

Co-authored-by: thisguymartin <13192083+thisguymartin@users.noreply.github.com>

Author: Copilot

CLAUDE.md

@@ -96,12 +96,12 @@ AI agent and assistant configs are included but optional -- the core workflow (t
 - `@planning-agent` -- architecture, task breakdown, and plan review (Gemini 3.1 Pro)
 - `build` mode in `opencode.json` -- default free implementation lane (MiniMax M2.5 Free)
 - `@engineer` -- premium implementation/orchestration for harder work (Gemini 3.1 Pro)
-- `@coder` -- autonomous test-fix loops when the spec and tests are clear (GPT 5.3 Codex)
+- `@coder` -- autonomous test-fix loops when the spec and tests are clear (GPT-4o via OpenAI API)
 - `@frontend` -- cost-sensitive UI and component work (Kimi K2.5)
 - `@reviewer` / `@security` -- code review and security audit (Gemini 3.1 Pro, cross-model)
-- `@qa` -- test generation, execution, and BDD flows (GPT 5.4 Mini)
+- `@qa` -- test generation, execution, and BDD flows (GPT-4o-mini via OpenAI API)
 - `@docs_generator` -- documentation (Gemini 3 Flash)
-- `@linear` -- Linear project management (GPT 5.4 Mini)
+- `@linear` -- Linear project management (GPT-4o-mini via OpenAI API)
 - `@pickle-think` -- free triage and rough planning (Big Pickle)
 - `@pickle-implement` -- free low-risk code changes (Big Pickle)
 

README.md

@@ -87,13 +87,13 @@ The repo includes configs for AI coding tools — these are entirely optional an
 | Agent | Specialty | Model | Cost | Invoke With |
 | :--- | :--- | :--- | :--- | :--- |
 | **Architect** | System design, task breakdown, plan review | Gemini 3.1 Pro | $2/$12 | `@planning-agent` |
-| **PM** | Linear integration | GPT 5.4 Mini | $0.75/$4.50 | `@linear` |
+| **PM** | Linear integration | GPT-4o-mini (OpenAI) | $0.15/$0.60 | `@linear` |
 | **Engineer** | Premium builder (approval) | Gemini 3.1 Pro | $2/$12 | `@engineer` |
-| **Coder** | Autonomous test-fix loops | GPT 5.3 Codex | $1.75/$14 | `@coder` |
+| **Coder** | Autonomous test-fix loops | GPT-4o (OpenAI) | $2.50/$10 | `@coder` |
 | **Frontend** | UI/component development | Kimi K2.5 | $0.60/$3 | `@frontend` |
 | **Reviewer** | Code quality review | Gemini 3.1 Pro | $2/$12 | `@reviewer` |
 | **Security** | Vulnerability scanning | Gemini 3.1 Pro | $2/$12 | `@security` |
-| **QA** | Test generation, execution, BDD | GPT 5.4 Mini | $0.75/$4.50 | `@qa` |
+| **QA** | Test generation, execution, BDD | GPT-4o-mini (OpenAI) | $0.15/$0.60 | `@qa` |
 | **Docs** | Inline + external docs | Gemini 3 Flash | ~$0.50/$3 | `@docs_generator` |
 | **Pickle Think** | Free triage & rough planning | Big Pickle | FREE | `@pickle-think` |
 | **Pickle Implement** | Free low-risk code changes | Big Pickle | FREE | `@pickle-implement` |

opencode/aig_agents/README.md

@@ -11,13 +11,13 @@ A consolidated 11-agent OpenCode setup tuned around Zen pricing and current mode
 | Agent | Purpose | Model | Cost (per 1M in/out) | Invocation |
 |-------|---------|-------|----------------------|------------|
 | `planning-agent` | Architecture design, task breakdown, plan review | Gemini 3.1 Pro | $2/$12 | `@planning-agent` |
-| `linear` | Create Linear projects/issues | GPT 5.4 Mini | $0.75/$4.50 | `@linear` |
+| `linear` | Create Linear projects/issues | GPT-4o-mini (OpenAI) | $0.15/$0.60 | `@linear` |
 | `engineer` | Premium senior builder (approval required) | Gemini 3.1 Pro | $2/$12 | `@engineer` |
-| `coder` | Autonomous test-fix loops | GPT 5.3 Codex | $1.75/$14 | `@coder` |
+| `coder` | Autonomous test-fix loops | GPT-4o (OpenAI) | $2.50/$10 | `@coder` |
 | `frontend` | UI/component development | Kimi K2.5 | $0.60/$3 | `@frontend` |
 | `reviewer` | Code quality review | Gemini 3.1 Pro | $2/$12 | `@reviewer` |
 | `security` | Security audit | Gemini 3.1 Pro | $2/$12 | `@security` |
-| `qa` | Test generation, execution, BDD | GPT 5.4 Mini | $0.75/$4.50 | `@qa` |
+| `qa` | Test generation, execution, BDD | GPT-4o-mini (OpenAI) | $0.15/$0.60 | `@qa` |
 | `docs_generator` | Documentation (inline + external) | Gemini 3 Flash | ~$0.50/$3 | `@docs_generator` |
 | `pickle-think` | Free triage, brainstorming, rough planning | Big Pickle | Free | `@pickle-think` |
 | `pickle-implement` | Free low-risk implementation | Big Pickle | Free | `@pickle-implement` |
@@ -33,8 +33,8 @@ A consolidated 11-agent OpenCode setup tuned around Zen pricing and current mode
 | **Premium Review** | Gemini 3.1 Pro | $2/$12 | planning-agent, reviewer, security |
 | **Premium Builder** | Gemini 3.1 Pro | $2/$12 | engineer |
 | **Builder Default** | MiniMax M2.5 Free | Free | default build mode in `opencode.json` |
-| **Specialist** | GPT 5.3 Codex | $1.75/$14 | coder |
-| **Workhorse** | GPT 5.4 Mini | $0.75/$4.50 | qa, linear |
+| **Specialist** | GPT-4o (OpenAI) | $2.50/$10 | coder |
+| **Workhorse** | GPT-4o-mini (OpenAI) | $0.15/$0.60 | qa, linear |
 | **Frontend** | Kimi K2.5 | $0.60/$3 | frontend |
 | **Budget** | Gemini 3 Flash | ~$0.50/$3 | docs_generator |
 | **Free** | Big Pickle | Free | pickle-think, pickle-implement |
@@ -64,7 +64,7 @@ A consolidated 11-agent OpenCode setup tuned around Zen pricing and current mode
 
 ### 2. Linear (`linear.md`)
 
-**Model:** GPT 5.4 Mini
+**Model:** GPT-4o-mini (OpenAI API)
 
 **Purpose:** Create Linear projects and issues with better structured output than the cheaper routing tier, while still staying in the low-cost bucket.
 
@@ -88,7 +88,7 @@ A consolidated 11-agent OpenCode setup tuned around Zen pricing and current mode
 
 ### 4. Coder (`coder.md`)
 
-**Model:** GPT 5.3 Codex
+**Model:** GPT-4o (OpenAI API)
 
 **Purpose:** Autonomous spec-driven implementation with test-fix loops. This is the right agent when the work is well-scoped and success is measured by green tests.
 
@@ -140,7 +140,7 @@ A consolidated 11-agent OpenCode setup tuned around Zen pricing and current mode
 
 ### 8. QA (`qa.md`)
 
-**Model:** GPT 5.4 Mini
+**Model:** GPT-4o-mini (OpenAI API)
 
 **Purpose:** Test generation and execution across languages, plus BDD / requirements-driven test generation when the user starts from behavior instead of code.
 
@@ -242,8 +242,8 @@ A consolidated 11-agent OpenCode setup tuned around Zen pricing and current mode
 | Criteria | `engineer` | `coder` | `frontend` |
 |----------|-----------|---------|-----------|
 | **Use when** | Highest-stakes implementation | Spec + tests are clear | UI/component work |
-| **Model** | Gemini 3.1 Pro | GPT 5.3 Codex | Kimi K2.5 |
-| **Cost** | $2/$12 | $1.75/$14 | $0.60/$3 |
+| **Model** | Gemini 3.1 Pro | GPT-4o (OpenAI) | Kimi K2.5 |
+| **Cost** | $2/$12 | $2.50/$10 | $0.60/$3 |
 | **Approval** | Requires approval | Autonomous loops | Full access |
 | **Best at** | Hard trade-offs, risky implementation | Test-fix cycles | Responsive UI, a11y |
 

opencode/aig_agents/coder.md

@@ -1,7 +1,7 @@
 ---
 description: Autonomous Coder - Spec-Driven Test-Fix Loops
 mode: subagent
-model: opencode/gpt-5.3-codex
+model: gpt-4o
 temperature: 0.1
 tools:
   read: true
@@ -14,7 +14,7 @@ tools:
 
 You are an **Autonomous Implementation Agent** trained for test-driven development loops. You take a specification and existing tests, then iterate: implement, run tests, fix, repeat — until all tests pass. Minimal human intervention.
 
-GPT 5.3 Codex is RL-trained for agentic coding — precise, spec-driven, terminal-optimized. Use it when the job is to close the loop from spec to green tests with minimal drift.
+GPT-4o is OpenAI's latest and most capable model — precise, spec-driven, and optimized for coding tasks. Use it when the job is to close the loop from spec to green tests with minimal drift.
 
 ---
 

opencode/aig_agents/linear.md

@@ -1,7 +1,7 @@
 ---
 description: Linear Project & Issue Creator via MCP
 mode: subagent
-model: opencode/gpt-5.4-mini
+model: gpt-4o-mini
 temperature: 0.2
 # tools:
 #   mcp: linear

opencode/aig_agents/qa.md

@@ -1,7 +1,7 @@
 ---
 description: QA Automation Architect (Multi-Language Test Coverage)
 mode: subagent
-model: opencode/gpt-5.4-mini
+model: gpt-4o-mini
 temperature: 0.1
 tools:
   bash: true