benchmark evidence
BFCL Non-Live
BFCL non-live function calling accuracy.
winner on BFCL Non-Live
direct benchmark result, not a broad vertical composite | source row dated 2026-05-15
scored on 2026-05-15
latest mapped results | top 20
| # | Model | Score | Evidence | Tested |
|---|---|---|---|---|
| 1 | Mistral: Mistral Small 4 | 89.7 | model-only independent_benchmark | 2026-05-15 |
| 2 | Anthropic: Claude Opus 4.5 | 89.7 | model-only independent_benchmark | 2026-05-15 |
| 3 | Meta: Llama 4 Scout | 89.4 | model-only independent_benchmark | 2026-05-15 |
| 4 | OpenAI: GPT-4.1 | 88.7 | model-only independent_benchmark | 2026-05-15 |
| 5 | Meta: Llama 4 Maverick | 88.7 | model-only independent_benchmark | 2026-05-15 |
| 6 | Google: Gemini 2.5 Flash | 88.1 | model-only independent_benchmark | 2026-05-15 |
| 7 | Meta: Llama 3.3 70B Instruct | 88.0 | model-only independent_benchmark | 2026-05-15 |
| 8 | Google: Gemma 3 27B | 87.2 | model-only independent_benchmark | 2026-05-15 |
| 9 | Amazon: Nova Pro 1.0 | 86.6 | model-only independent_benchmark | 2026-05-15 |
| 10 | Mistral Large | 83.0 | model-only independent_benchmark | 2026-05-15 |
| 11 | MoonshotAI: Kimi K2 Thinking | 81.6 | model-only independent_benchmark | 2026-05-15 |
| 12 | OpenAI: o4 Mini | 81.3 | model-only independent_benchmark | 2026-05-15 |
| 13 | OpenAI: GPT-5.2 | 78.3 | model-only independent_benchmark | 2026-05-15 |
| 14 | Anthropic: Claude Sonnet 4.5 | 59.8 | model-only independent_benchmark | 2026-05-15 |
| 15 | Anthropic: Claude Haiku 4.5 | 55.4 | model-only independent_benchmark | 2026-05-15 |
| 16 | OpenAI: o3 | 40.4 | model-only independent_benchmark | 2026-05-15 |
| 17 | DeepSeek: DeepSeek V3.2 | 34.9 | model-only independent_benchmark | 2026-05-15 |
what this result means
BFCL non-live function calling accuracy.
This benchmark contributes direct public evidence. Read its scope before generalizing the result.
A win here is a win on BFCL Non-Live. Broad task pages require independent corroboration before naming a general winner.
source record
category: structured_output
metric: accuracy
matched models: 17
latest source date: 2026-05-15
direction: higher is better