← Back to leaderboard

mistral:7b

5.6 GB measured VRAM · 4-8GB tier · tool-gauntlet-v1
Tier 1 — floor 5/5
seed 1 (passed): correct single call
21 tokens · 6.499s
assistant
get_weather({'city': 'Seattle'})
Tier 2 — restraint 0/5
seed 1 (FAILED): called get_weather when no tool was needed
126 tokens · 0.837s
assistant
get_weather({'city': 'Tokyo'})
get_weather({'city': 'Tokyo'})
Tier 3 — selection 4/5
seed 1 (FAILED): expected exactly 1 tool call, got 3
147 tokens · 1.016s
assistant
get_time({'city': 'Tokyo'})
get_weather({'city': 'New York City'})
convert_currency({'amount': 100, 'from_currency': 'USD', 'to_currency': 'EUR'})
Tier 4 — extraction 5/5
seed 1 (passed): correctly extracted and typed all fields from messy prose
88 tokens · 0.617s
assistant
book_flight({'date': '2026-03-24', 'passengers': 3, 'origin': 'Chicago', 'destination': 'Denver'})
Tier 5 — chaining 0/5
seed 1 (FAILED): first turn: did not call get_user_location
97 tokens · 0.675s
assistant
Tier 6 — strict schema 5/5
seed 1 (passed): correctly extracted enum, array, and date under full schema constraints
92 tokens · 0.638s
assistant
create_calendar_event({'date': '2026-04-02', 'priority': 'high', 'attendees': ['Maria', 'James'], 'title': 'Budget Review'})