dottxt Tool-call Index: same open model's tool-call failure varies 13x across providers

dottxt's Tool-call Index shows the same open-weight model's tool-call failure rate varies up to 13-14x across inference providers, with stepwise jumps in failure rates matching providers' silent inference-stack updates.

2026-11-11 ~ 2026-11-12 · 2 related posts

• dottxt launches Tool-call index: same open model fails tool calls up to 13x more on some providers — remilouf · 2026-11-11
• Tool-call failure rates jump 14x across providers, charts pinpoint inference stack updates — remilouf · 2026-11-12