of DeepSeek V4 Flash 0731's tool calls fail on Alibaba Cloud Int.
Updated October 29, 2026
Tool-call index
Not all models are served equally by all providers. On a typical model, the worst provider breaks tool calls 19× as often as the best. Find the best one for your agents.
| # | Model | Broken tool calls | Rank | Best on the model | Tool calls a day |
|---|---|---|---|---|---|
| 1 | GLM 5.1fp8 | 0.05% | 2 of 7 | Friendli 0.00% | 1.1k |
| 2 | Kimi K3 | 0.11% | 11 of 24 | DeepInfra (mxfp4) 0.01% | 77k |
| 3 | DeepSeek V4.1 Flash 1017 | 0.57% | 18 of 35 | Sail Research (fp4) 0.03% | 90k |
| 4 | Qwen3.5 397B A17B | 0.58% | 1 of 7 | — | 6.6k |
| 5 | DeepSeek V4 Pro 0813 | 0.74% | 8 of 21 | Ionstream 0.04% | 32k |
| 6 | DeepSeek V4.1 Flash | 0.79% | 22 of 35 | Prime Intellect 0.06% | 160k |
| 7 | Qwen3.8 2.4T A95B | 1.33% | 3 of 5 | SiliconFlow (fp8) 0.19% | 3.3k |
| 8 | Kimi K2.7 Codefp8 | 1.35% | 9 of 11 | NovitaAI (int4) 0.03% | 1.2k |
| 9 | GLM 5.3 | 1.82% | 32 of 42 | Amazon Bedrock 0.08% | 10k |
| 10 | GLM 5.2fp8 | 2.18% | 24 of 27 | Friendli 0.15% | 4.8k |
| 11 | Qwen3.8 27B | 2.44% | 19 of 21 | Darkbloom (fp4) 0.07% | 84k |
| 12 | GLM 5.2fp8/fast | 2.63% | 26 of 27 | Friendli 0.15% | 3.6k |
| 13 | DeepSeek V4 Flash 0423fp8 | 2.70% | 11 of 16 | Crusoe (fp8) 0.46% | 43k |
| 14 | GLM 5.3prime | 2.85% | 37 of 42 | Amazon Bedrock 0.08% | 7.7k |
| 15 | DeepSeek V4 Pro 0423fp8 | 3.32% | 9 of 13 | Cloudflare 0.78% | 39k |
| 16 | DeepSeek V4 Flash 0731 | 15.01% | 22 of 22 | Sail Research (fp4) 0.04% | 24k |
Get updates when a new model drops
And which provider to use.
For inference providers
Tool calling doesn't have to break
The same provider can handle tool calls well on one model and struggle on the next.
on GLM 5.1, its best model.
models, where the settings tell OpenRouter to skip Alibaba Cloud Int..