of gpt-oss-120b's tool calls fail on Nebius Token Factory (fp4).
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 | Muse Spark Openfp4 | 0.07% | 1 of 6 | — | 13k |
| 2 | DeepSeek V4.1 Flashfp8 | 0.28% | 5 of 35 | Prime Intellect 0.06% | 35k |
| 3 | DeepSeek V4 Pro 0813fp8 | 0.35% | 4 of 21 | Ionstream 0.04% | 1.2k |
| 4 | GLM 5.3 Flashfp4 | 0.38% | 10 of 34 | NextBit (fp4) 0.00% | 12k |
| 5 | GLM 5.3fp4 | 0.75% | 20 of 42 | Amazon Bedrock 0.08% | 3.7k |
| 6 | GLM 5.2fp4 | 1.02% | 15 of 27 | Friendli 0.15% | 12k |
| 7 | Kimi K2.7 Codefp4 | 1.28% | 8 of 11 | NovitaAI (int4) 0.03% | 3.2k |
| 8 | Nemotron 3.5 Super VLbf16 | 1.34% | 3 of 8 | Together 0.66% | 4.0k |
| 9 | Beam 1.0 501B A23Bfp4 | 1.35% | 2 of 3 | Together 1.01% | 142k |
| 10 | DeepSeek V4.1 Flash 1017fp8 | 1.46% | 24 of 35 | Sail Research (fp4) 0.03% | 107k |
| 11 | Qwen3 30B A3B Instruct 2507fp8 | 2.89% | 1 of 2 | — | 25k |
| 12 | gpt-oss-120bfp4 | 80.12% | 15 of 15 | Groq 0.04% | 3.9k |
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 Muse Spark Open, its best model.
models, where the settings tell OpenRouter to skip Nebius Token Factory.