chat model, one letter · New models: first look (50 smoke positions)
Mercury 3low
Smoke test
Results f24d1f41, updated 16 November 2026, 22:20 UTC; scored by GNU Backgammon 1.08, 2-ply.
How we called it
- Model ID
- inception/mercury-3
- Provider or endpoint
- OpenRouter, pinned to Inception (its only endpoint; quantization not published)
- Date run
- 2026-10-21
- Price
- $0.1 / $0.35 per million tokens in / out; API price; price date 2026-10-21
- Model type
- chat model, one letter
- Weights
- closed
- Runs on
- API
- Reasoning setting
- reasoning effort low
- Confidence source
- none
- Rules text
- long
- Option groups
- one request (no split)
- Sampling settings
- temperature 1.0
- Notes
- unconstrained letter; reasoning text never returned
Smoke test
- Positions answered
- 50 of 50 (0 failed, 0 timed out)
- Invalid answers
- 5
- Error rate (PR)
- 23.6 (95% interval 10.5 to 39.0)
- Median / P90 request s
- 23.0 / 78.7
- $ per 1,000 decisions
- 4.85