Issues

• [Bug] Streamed Gemini (google-genai) calls lose finish_reason, and chunks without content or usage are dropped from the trace

  wandb/weave#8041 · Filter by author pcbeingused333 opened on Oct 16, 2026
  1

• [Bug] Streamed Groq completions lose tool calls, reasoning and finish_reason (AttributeError when the first chunk has a tool call)

  wandb/weave#8039 · Filter by author pcbeingused333 opened on Oct 16, 2026

• [Bug] Streamed Anthropic responses lose thinking and tool_use blocks in the trace

  wandb/weave#8037 · Filter by author pcbeingused333 opened on Oct 16, 2026

• [Bug]: LangChain integration bills prompt-cache tokens at the full input rate

  wandb/weave#8031 · Filter by author chrikrah opened on Oct 14, 2026

• [Bug]: Importing weave changes the process-wide locale

  bug
  wandb/weave#8028 · Filter by author NiklasvonM opened on Oct 14, 2026

• [Bug]: Google GenAI thinking tokens are left out of the completion cost

  wandb/weave#8017 · Filter by author chrikrah opened on Oct 11, 2026
  1

• Evaluation summaries overwrite sibling scores when nested keys repeat

  wandb/weave#7975 · Filter by author March-7 opened on Oct 4, 2026

• Bedrock: GPT-6 Astra calls made with an inference profile ARN get no cost

  wandb/weave#7943 · Filter by author kimnamu opened on Sep 24, 2026

• StringMatchScorer scores an empty output as a match against every target

  wandb/weave#7939 · Filter by author shaurya416 opened on Sep 23, 2026

• Add OrcaRouter as an optional AI provider

  wandb/weave#7832 · Filter by author putraperdana1207-pixel opened on Sep 3, 2026

• trace_usage double-counts client-rolled-up summaries in single-process traces

  wandb/weave#7831 · Filter by author roy-tong opened on Sep 3, 2026
  2

• OpenAI-compat: weave.init then OpenAI base_url /v1, not origin-only local runners

  wandb/weave#7812 · Filter by author cursor opened on Aug 28, 2026

• EvalPort import/export for Weave Datasets and Evaluation results (adapter already built and shipped)

  wandb/weave#7750 · Filter by author adhabnr-ux opened on Aug 14, 2026

• [Bug]: EvaluationLogger.log_summary(summary=..., auto_summarize=False) duplicates the custom summary in the logged output

  bug
  wandb/weave#7741 · Filter by author qklopfenstein-owkin opened on Aug 13, 2026

• OpenAI streaming silently not traced when extra_headers=None is passed explicitly

  wandb/weave#7731 · Filter by author Shailendra005 opened on Aug 12, 2026

• Renovate Dashboard

  wandb/weave#7395 · Filter by author wandb-renovate opened on Jun 25, 2026

• Bedrock Nova models in ap-* regions get an unmatched "ap." prefix instead of "apac."

  wandb/weave#7325 · Filter by author kimnamu opened on Jun 19, 2026

• [Feature] Span-level observability for LLM RAG tracing — per-retrieval-hop quality scores and chain degradation attribution

  wandb/weave#7280 · Filter by author Ruthwik-Data opened on Jun 17, 2026

• [Bug]: OpenTelemetry ingestion results in unusable threads.

  bug
  wandb/weave#7066 · Filter by author a0js opened on Jun 5, 2026
  1

• Add HVTracker badge to README?

  wandb/weave#7014 · Filter by author YugantM opened on May 30, 2026

• Feature: span-level evaluation metrics for RAG pipelines in Weave traces

  enhancement
  wandb/weave#6946 · Filter by author Ruthwik-Data opened on May 22, 2026
  2

• [Bug]: 429 handling in Node SDK doesn't distinguish WAIT / CAP / STOP — retry amplification risk

  bug
  wandb/weave#6543 · Filter by author SirBrenton opened on Apr 2, 2026
  1

• [Q]: Why can't wandb.ai access?

  question
  wandb/weave#6331 · Filter by author soulcape opened on Mar 12, 2026
  2

• [Bug]: It looks like weave demo shows a panel per run but when run it does it per project

  bug
  wandb/weave#5962 · Filter by author jwong8314 opened on Jan 22, 2026

• [Bug]: Annoying bug on adding feedbacks in weave evaluation

  bug
  wandb/weave#5926 · Filter by author sanggusti opened on Jan 20, 2026