Granite 4.0 H 350M: benchmark results
IBM's Granite 4.0 H 350M instruct model, fine-tuned from Granite-4.0-H-350M-Base with supervised fine-tuning, reinforcement learning and model merging, part of the Granite 4.0 Nano collection. Provider: IBM. Released 2025-10-28. Access: Open.
Aggregate Rating 1266 ± 28, rank #1502 of 1635 rated models, from 38 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Kurakurai French LLM - GPQA Diamond | 26.92 | Accuracy (%) | 61 |
| Kurakurai French LLM - BFCL v2 | 55.38 | Accuracy (%) | 48.3 |
| Kurakurai French LLM - HumanEval+ | 24 | pass@1 (%) | 43.9 |
| Kurakurai French LLM - MBPP | 31.98 | pass@1 (%) | 43.9 |
| AA Humanity's Last Exam | 6.44 | Accuracy (%) | 37.7 |
| AA CritPt | 0 | Accuracy (%) | 20.9 |
| Kurakurai French LLM - MATH-500 | 9.5 | Accuracy (%) | 17.1 |
| Kurakurai French LLM - AIME 2025 | 0 | Accuracy (%) | 15.9 |
| Kurakurai French LLM - Global MMLU | 28.7 | Accuracy (%) | 14.6 |
| Kurakurai French LLM - IFEval | 37.23 | Accuracy (%) | 14.6 |
| Kurakurai French LLM - MMLU-Pro | 9.13 | Accuracy (%) | 14.6 |
| AA Omniscience - Software Engineering (SWE) - Kotlin | 8 | Accuracy (%) | 13.7 |
Interactive version: theaggregate.ai/model?slug=granite-4-0-h-350m · How It Works · Data refreshed daily, snapshot 2026-11-05.