OpenCode Test
|
f63172c4cf
|
feat(external-llm): standardize tiers and optimize model selection
- Rename tiers: opus/sonnet/haiku → frontier/mid-tier/lightweight
- Align with industry benchmarks (MMLU, GPQA, Chatbot Arena)
- Add /external command for LLM mode control
- Fix invoke.py timeout passthrough (now 600s default)
Tier changes:
- Promote gemini-2.5-pro to frontier (benchmark-validated)
- Demote glm-4.7 to mid-tier then removed (unreliable)
- Promote gemini-2.5-flash to mid-tier
New models added:
- gpt-5-mini, gpt-5-nano (GPT family coverage)
- grok-code (Grok/X family)
- glm-4.5-air (lightweight GLM)
Removed (redundant/unreliable):
- o3 (not available)
- glm-4.7 (timeouts)
- gpt-4o, big-pickle, glm-4.5-flash (redundant)
Final: 11 models across 3 tiers, 4 model families
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
|
2026-01-12 03:30:51 -08:00 |
|
OpenCode Test
|
df6cf94dae
|
feat(external-llm): add external LLM integration (fc-004)
Implements external LLM routing via opencode CLI for:
- GitHub Copilot (gpt-5.2, claude-sonnet-4.5, claude-haiku-4.5, o3, gemini-3-pro)
- Z.AI (glm-4.7 for code generation)
- OpenCode native (big-pickle)
Components:
- mcp/llm-router/invoke.py: Main router with task-based model selection
- mcp/llm-router/delegate.py: Agent delegation helper (respects external mode)
- mcp/llm-router/toggle.py: Enable/disable external-only mode
- mcp/llm-router/providers/: CLI wrappers for opencode and gemini
Features:
- Persistent toggle via state/external-mode.json
- Task routing: reasoning -> gpt-5.2, code-gen -> glm-4.7, long-context -> gemini
- Claude tier mapping: opus -> gpt-5.2, sonnet -> claude-sonnet-4.5, haiku -> claude-haiku-4.5
- Session-start hook announces when external mode is active
- Natural language toggle support via component registry
Plan: gleaming-routing-mercury
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
|
2026-01-08 13:34:35 -08:00 |
|