Best LLMs for Tool Calling
The best LLM for tool calling is GLM 5.2, completing 99.1% of Tau²-Bench's multi-turn tool-use tasks. GLM 4.7 Flash (98.8%) and Claude Fable 5 (98.5%) round out the top three.
- 1
glm-5.2ReasoningToolsJSON51.1 intel · $0.760/M · 143 t/s99.1%τ²-Bench - 2glm-4.7-flashReasoningToolsJSON22.9 intel · $0.060/M · 203K ctx98.8%τ²-Bench
- 3
claude-fable-5ReasoningToolsJSON+159.9 intel · $10.00/M · 64 t/s98.5%τ²-Bench - 4
step-3.7-flashReasoningToolsJSON+130.3 intel · $0.200/M · 157 t/s98.5%τ²-Bench - 5glm-5v-turboReasoningToolsJSON+134.5 intel · $1.20/M · 203K ctx98.5%τ²-Bench
- 6glm-5-turboReasoningToolsJSON38.1 intel · $1.20/M · 203K ctx98.5%τ²-Bench
- 7glm-5ReasoningToolsJSON39.5 intel · $0.950/M · 205K ctx98.2%τ²-Bench
- 8
grok-4.3ReasoningToolsJSON+137.6 intel · $1.25/M · 1M ctx97.7%τ²-Bench - 9glm-5.1ReasoningToolsJSON40.2 intel · $0.966/M · 205K ctx97.7%τ²-Bench
- 10
qwen3.6-plusReasoningToolsJSON+139.6 intel · $0.325/M · 55 t/s97.7%τ²-Bench - 11
deepseek-v4-proReasoningToolsJSON44.3 intel · $0.435/M · 67 t/s96.2%τ²-Bench - 12qwen3.6-max-previewReasoningToolsJSON40.0 intel · $1.03/M · 262K ctx95.9%τ²-Bench
- 13
kimi-k2.6ReasoningToolsJSON+144.2 intel · $0.589/M · 262K ctx95.9%τ²-Bench - 14kimi-k2.5ReasoningToolsJSON+135.4 intel · $0.570/M · 262K ctx95.9%τ²-Bench
- 15glm-4.7ReasoningToolsJSON33.7 intel · $0.400/M · 541 t/s95.9%τ²-Bench
- 16
gemini-3.1-pro-previewReasoningToolsJSON+246.5 intel · $2.00/M · 136 t/s95.6%τ²-Bench - 17qwen3.5-397b-a17bReasoningToolsJSON+133.7 intel · $0.390/M · 69 t/s95.6%τ²-Bench
- 18gemini-3.5-flashReasoningToolsJSON+250.2 intel · $1.50/M · 275 t/s95.3%τ²-Bench
- 19qwen3.6-35b-a3bReasoningToolsJSON+131.6 intel · $0.140/M · 158 t/s95.3%τ²-Bench
- 20
minimax-m2.5ReasoningToolsJSON33.7 intel · $0.150/M · 205K ctx95.3%τ²-Bench - 21qwen3.7-maxReasoningToolsJSON46.0 intel · $1.48/M · 45 t/s94.7%τ²-Bench
- 22claude-opus-4.8ReasoningToolsJSON+155.7 intel · $5.00/M · 77 t/s94.4%τ²-Bench
- 23
mistral-medium-3-5ReasoningToolsJSON+129.9 intel · $1.50/M · 152 t/s94.2%τ²-Bench - 24qwen3.6-27bReasoningToolsJSON+137.1 intel · $0.289/M · 60 t/s94.2%τ²-Bench
- 25
mimo-v2.5-proReasoningToolsJSON42.2 intel · $0.435/M · 65 t/s94.2%τ²-Bench
How this is ranked
AI models ranked by Tau²-Bench — multi-turn conversations where the model has to call the right tools, in the right order, against a real API to complete a customer task. This measures whether function calling actually works under pressure, which is a different question from whether a model supports the parameter at all.
Frequently asked
Which LLM is best at tool calling?
The best LLM for tool calling is GLM 5.2, completing 99.1% of Tau²-Bench's multi-turn tool-use tasks. GLM 4.7 Flash (98.8%) and Claude Fable 5 (98.5%) round out the top three.
Which AI model is best at function calling?
The best AI model for tool calling is GLM 5.2, completing 99.1% of Tau²-Bench's multi-turn tool-use tasks. GLM 4.7 Flash (98.8%) and Claude Fable 5 (98.5%) round out the top three.
What's a good alternative to GLM 5.2?
GLM 4.7 Flash (98.8%) is the closest alternative on this metric, followed by Claude Fable 5 (98.5%). See the full ranking above for the tradeoffs.