Gemini vision (solves chess), solver plain-text fallback, pandas in python_repl 0c9c31a DriptoBhattacharyya Claude Opus 4.8 commited on Jun 2
Fetch task files from HF Hub GAIA dataset (scoring API /files is 404) 7f8f0e0 DriptoBhattacharyya Claude Opus 4.8 commited on Jun 2
Pin mistral-large backend, add web/youtube/python-exec tools, guard None structured outputs 00d7c26 DriptoBhattacharyya Claude Opus 4.8 commited on Jun 2
Cache good answers by task_id so re-runs skip solved questions ef849ff DriptoBhattacharyya Claude Opus 4.8 commited on Jun 2
Raise default QUESTION_TIMEOUT to 500s for gateway stalls 21bbba7 DriptoBhattacharyya Claude Opus 4.8 commited on Jun 2
Slim graph + cap tool rounds for gateway latency; pin gpt-oss-120b 6fe947e DriptoBhattacharyya Claude Opus 4.8 commited on Jun 2
Add openai_compatible provider branch for freellmapi gateway 37bc9aa DriptoBhattacharyya Claude Opus 4.8 commited on Jun 2
Default to Gemini 2.5-flash (thinking disabled) for text/judge nodes abd9bdb DriptoBhattacharyya Claude Opus 4.8 commited on Jun 2
Add provider toggle (Groq/Gemini) + global rate limiter to beat free-tier limits 86a7fd1 DriptoBhattacharyya Claude Opus 4.8 commited on Jun 2
Harden agent: survive Groq tool_use_failed, enforce per-question timeout 0a9b8d0 DriptoBhattacharyya Claude Opus 4.8 commited on Jun 2
GAIA LangGraph agent: Groq + Tavily + LLM-as-judge cb1b811 DriptoBhattacharyya Claude Opus 4.8 commited on Jun 2