xAI·released 2025-04-091 source
Grok-3 mini benchmark scores: 10 benchmarks tracked. 40% of its scores are independently reproduced — the rest are self-reported or unverified.
Leads on 0 of 10 benchmarks·best result 90.94 on MATH level 5 (reproduced)·4 of 10 independently reproduced·$0.3/$0.5 per M tokens
Consensus: LiteLLM · See every model’s pricing →
| Benchmark | Score | Evidence status | Measured |
|---|---|---|---|
| MATH level 5 | 90.94accuracy (%) | reproduced· optimizedT1 | 2025-04-09Epoch AI |
| OTIS Mock AIME 2024-2025 | 77.76accuracy (%) | reproduced· optimizedT1 | 2025-04-09Epoch AI |
| Lech Mazur Writing | 73.5accuracy (%) | unverifiedT1 | 2025-04-09lechmazur/writing Github repository |
| GPQA diamond | 68.35accuracy (%) | reproduced· optimizedT1 | 2025-04-09Epoch AI |
| Fiction.LiveBench | 66.7accuracy (%) | unverifiedT1 | 2025-04-09Fiction.live leaderboard |
| Aider polyglot | 49.3accuracy (%) | unverified· optimizedT1 | 2025-04-09https://aider.chat/docs/leaderboards/ |
| WeirdML | 42.58accuracy (%) | unverifiedT1 | 2025-06-24WeirdML Leaderboard |
| ARC-AGI | 16.5accuracy (%) | unverified· optimizedT1 | 2025-06-24ARC Prize Leaderboard |
| FrontierMath-2025-02-28-Private | 10.28accuracy (%) | reproduced· optimizedT1 | 2025-04-09Epoch AI |
| ARC-AGI-2 | 0.42accuracy (%) | unverifiedT2 | 2025-06-24 |
Claims drawn from cited facts, not live model generation.
This model originates from xAI, the developer credited with its creation. It dates to 2025, placing its vintage in the current generation of models.
2 cited facts
This model is scored on 10 tracked benchmarks, and it currently holds no top score on any of them. Its record has been independently reproduced, meaning outside evaluation backs the reported scores up rather than resting on vendor claims alone.
3 cited facts
Under the disclosed evaluation harness, this model trails the state-of-the-art leader by an average of 42.22 points, a wide margin that reads as well behind the leaders rather than at the frontier. It holds the top score on 0 benchmarks, meaning there is no category leadership to claim on any of its measured tasks.
2 cited facts
6 facts cross-checked across data sources: 6 single-source.
Source facts, citations, and refresh stamp for this record.
Sources: epoch_benchmarks, litellm_prices
Grok-3 mini is an AI model developed by xAI, released 2025-04-09. Its benchmark record below tags every score as independently reproduced, self-reported, or unverified.
Grok-3 mini has recorded scores on 10 benchmarks, each shown with its evidence status.
Grok-3 mini has recorded scores on 10 benchmarks — Lech Mazur Writing, GPQA diamond, MATH level 5, OTIS Mock AIME 2024-2025, WeirdML, FrontierMath-2025-02-28-Private, and 4 more. The full table above shows each score with its evidence status.
4 of 10 recorded scores (40%) are independently reproduced rather than self-reported by the lab.
Grok-3 mini has 10 tracked claims: 4 independently reproduced, 6 unverified.
Listed API pricing: $0.3 per million input tokens, $0.5 per million output tokens. See the pricing block for the full breakdown.