List your company

Multi-Turn Tool Calling on BFCL v4 (val)

85Overall AccuracyRCTP-FT + RC-GRPO
-3.419.5542.565.45Feb 3, 2026

Evaluation results

MethodLinks
RCTP-FT + RC-GRPOBackbone=Qwen2.5-7B2026.028577.7888.2381.8191.3
RCTP-FT + GRPOBackbone=Qwen2.5-7B2026.0273.7566.6764.7177.2782.61
Opus-4.5Version=claude-opus-4-5-202511012026.0261.2555.5665.2264.7159.09
GLM-4.7Version=glm-4.72026.0258.7555.5658.8254.5565.22
Sonnet-4.5Version=claude-sonnet-4-5-202509292026.0257.544.4458.8259.0965.22
Gemini-3-ProVersion=gemini-3-pro-preview2026.0253.7555.5664.7136.3660.87
GPT-5.2Version=gpt-5.2-2025-12-112026.025038.8947.0645.4565.22
SFT + GRPOBackbone=Qwen2.5-7B2026.0248.7555.5635.295052.17
RCTP-FT + RC-GRPOBackbone=LLaMA-3.1-8B2026.0248.7538.8935.2960.8754.54
SFT + RC-GRPOBackbone=Qwen2.5-7B2026.0246.255035.295047.82
RCTP-FT + GRPOBackbone=LLaMA-3.1-8B2026.0246.2533.3335.2954.5556.52
SFT + GRPOBackbone=LLaMA-3.1-8B2026.023538.8917.6740.9139.13
SFT + RC-GRPOBackbone=LLaMA-3.1-8B2026.023544.4417.6436.3639.13
Base ModelBackbone=Qwen2.5-7B2026.0211.2511.1117.654.5413.04
Base ModelBackbone=LLaMA-3.1-8B2026.0200000