# Exp 5 Timeline

## Parameters
- **Architecture:** Single-agent (NO subagents)
- **Model:** glm-5.1 for all steps in main session
- **lightContext:** N/A (single session)
- **Start balance:** z.ai $8.54, Perplexity $29.13

## Hypothesis

**Expected cost:** ~$0.30 (только токены, без subagent overhead)
**Expected quality:** 8-10/10 (тестируем, падает ли качество без изоляции)

## Timeline

| Step | Start (UTC) | End (UTC) | Duration | Tokens | Notes |
|------|-------------|-----------|----------|--------|-------|
| 1. Planning | - | - | - | - | In main session |
| 2. Search | - | - | - | - | web_search × 8 |
| 3. Relevance | - | - | - | - | In main session |
| 4. Reading | - | - | - | - | web_fetch × N |
| 5. Gap Analysis | - | - | - | - | In main session |
| 6. Iterative Search | - | - | - | - | web_search × M |
| 7. Synthesis | - | - | - | - | In main session |

## Key Difference from Exp 3/4

| Factor | Exp 3/4 | Exp 5 |
|--------|---------|-------|
| Subagent spawns | 4 | **0** |
| Context isolation | Yes | No |
| Expected cost | ~$2.00 | **~$0.30** |
| Risk | Low | Quality may drop |