gpt-5.4-mini failing spatial-1; gemini-2.5-flash failing causality-1. gemini-2.5-flash scores rising.
July 25, 2026 — 8:39 PM CT
Drift Alerts
- SCORE_RISE gemini/gemini-2.5-flash on common-sense-1
Provider Status
- OpenAI Elevated errors affecting ChatGPT conversations
- OpenAI Elevated error rates
- OpenAI Elevated error rates
- OpenAI Elevated Error Rates
- OpenAI Elevated Errors in Codex Review
- Anthropic Elevated errors for Claude Fable 5, Claude Sonnet 5, Claude Haiku 4.5, and other models
- Anthropic Elevated errors for Mythos 5, Fable 5, Opus 5 and Claude Haiku 4.5
- Anthropic Sonnet 4.6 and Sonnet 5 errors elevated
- Anthropic Microsoft Office add-in availability
- Anthropic Elevated error rate on Opus 4.8
Scorecard
| Model | ambiguity-1 | causality-1 | code-1 | common-sense-1 | logic-1 | math-1 | spatial-1 |
|---|---|---|---|---|---|---|---|
| anthropic/claude-haiku-4-5 | ✓ (4.33) | ✓ (4.67) | ✓ (4.67) | ✓ (3.33) | ✓ (5) | ✓ (5) | ✓ (5) |
| anthropic/claude-opus-4-6 | ✓ (5) | ✓ (4.83) | ✓ (4.83) | ✓ (4.33) | ✓ (5) | ✓ (5) | ✓ (5) |
| anthropic/claude-sonnet-4-6 | ✓ (4.83) | ✓ (4.83) | ✓ (4.5) | ✓ (3.5) | ✓ (5) | ✓ (5) | ✓ (5) |
| gemini/gemini-2.5-flash | ✓ (4.5) | ✗ (3.5) | ✓ (4.67) | ✓ (4.83)was 3.83 | ✓ (4.83) | ✓ (5) | ✓ (5) |
| gemini/gemini-2.5-pro | ✓ (4.67) | ✓ (5) | ✓ (4.67) | ✓ (5) | ✓ (5) | ✓ (5) | ✓ (5) |
| ollama/llama3 | — | — | — | — | — | — | — |
| openai/gpt-5.4 | ✓ (4.33) | ✓ (5) | ✓ (4.67) | ✓ (4.33) | ✓ (5) | ✓ (4.33) | ✓ (5) |
| openai/gpt-5.4-mini | ✓ (4.5) | ✓ (4.67) | ✓ (4.67) | ✓ (4.5) | ✓ (5) | ✓ (5) | ✗ (3.67) |
Model Status
- → anthropic/claude-haiku-4-5 stable
- → anthropic/claude-opus-4-6 stable
- → anthropic/claude-sonnet-4-6 stable
- ↑ gemini/gemini-2.5-flash up
- → gemini/gemini-2.5-pro stable
- → openai/gpt-5.4 stable
- → openai/gpt-5.4-mini stable
Raw Data
- Detail log — full responses and judge verdicts per prompt
- JSON — structured data for programmatic access
- Markdown — plain text report
- responses.json — raw model outputs
- judgments.json — raw judge verdicts
- run.log — debug log
- Agent Skill — how to read and interpret this data
- Methodology — how evaluations work