of DeepSeek V4.1 Flash 1017's tool calls fail on GMICloud (fp8).
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.
| # | Provider | Broken tool calls | Tool calls a day | $ in / out per M | Tokens/s | Uptime |
|---|---|---|---|---|---|---|
| 1 | Sail Researchfp4# | 0.03% | 445k | $0.10 / $0.15 | 64 | 100.0% |
| 2 | Prime Intellect# | 0.04% | 311k | $0.30 / $1.20 | 207 | 100.0% |
| 3 | Ionstream# | 0.04% | 537k | $0.17 / $0.49 | 155 | 99.7% |
| 4 | Relace# | 0.10% | 2.8M | $0.07 / $0.50 | 78 | 100.0% |
| 5 | Morphfp8# | 0.11% | 2.1M | $0.04 / $0.38 | 101 | 99.8% |
| 6 | Wafermxfp4# | 0.20% | 2.4M | $0.05 / $0.50 | 214 | 100.0% |
| 7 | DeepInfrafp8/flex# | 0.27% | 317k | $0.08 / $0.24 | 50 | 84.4% |
| 8 | Fireworks Spotspot# | 0.28% | 1.7M | $0.21 / $0.84 | 127 | 100.0% |
| 9 | Open Inferencefp4# | 0.32% | 986k | $0.04 / $1.14 | 63 | 99.6% |
| 10 | Particle AI# | 0.34% | 3.4k | $0.25 / $1.00 | 189 | 100.0% |
| 11 | inference.netfp8# | 0.36% | 5.8M | $0.08 / $0.55 | 115 | 100.0% |
| 12 | Reka AI# | 0.46% | 125k | $0.06 / $0.60 | 93 | 97.3% |
| 13 | DigitalOcean# | 0.47% | 1.7M | $0.18 / $0.72 | 190 | 100.0% |
| 14 | DeepInfrafp8# | 0.49% | 2.0M | $0.10 / $0.29 | 55 | 99.7% |
| 15 | Parasailfp8# | 0.50% | 261k | $0.30 / $1.20 | 142 | 100.0% |
| 16 | Fireworks# | 0.51% | 156k | $0.30 / $1.20 | 114 | 98.7% |
| 17 | Cloudflare# | 0.57% | 537k | $0.10 / $0.60 | 94 | 99.9% |
| 18 | Alibaba Cloud Int.# | 0.57% | 90k | $0.17 / $0.66 | 105 | 99.4% |
| 19 | Modal# | 0.67% | 364k | $0.25 / $1.00 | 139 | 99.8% |
| 20 | DeepInfrafp8/priority# | 0.71% | 961k | $0.15 / $0.44 | 65 | 98.7% |
| 21 | Cohere# | 0.79% | 15k | $0.20 / $0.66 | 201 | 100.0% |
| 22 | Makorafp8# | 0.97% | 202k | $0.15 / $1.20 | 152 | 99.8% |
| 23 | StreamLakefp8# | 1.45% | 26k | $0.15 / $0.60 | 89 | 100.0% |
| 24 | Nebius Token Factoryfp8# | 1.46% | 107k | $0.30 / $1.20 | 142 | 99.4% |
| 25 | Crusoefp8# | 1.46% | 313k | $0.29 / $1.20 | 74 | 100.0% |
| 26 | Decartfp4# | 1.47% | 314k | $0.12 / $0.49 | 96 | 99.1% |
| 27 | SiliconFlowfp8# | 1.62% | 286k | $0.30 / $1.20 | 168 | 99.6% |
| 28 | Baidu Qianfanfp8# | 1.85% | 15k | $0.15 / $0.60 | 102 | 99.7% |
| 29 | Phala# | 1.92% | 64k | $0.20 / $1.20 | 108 | 99.7% |
| 30 | DeepSeek# | 2.01% | 11.4M | $0.15 / $0.60 | 77 | 100.0% |
| 31 | Baseten# | 2.57% | 713k | $0.30 / $1.20 | 193 | 100.0% |
| 32 | Together# | 3.44% | 5.8M | $0.30 / $1.20 | 250 | 100.0% |
| 33 | NovitaAIfp8# | 3.50% | 131k | $0.20 / $0.78 | 111 | 99.9% |
| 34 | AtlasCloudfp8# | 3.67% | 231k | $0.20 / $0.78 | 109 | 99.6% |
| 35 | GMICloudfp8# | 6.88% | 60k | $0.18 / $0.72 | 74 | 100.0% |
Recommended OpenRouter settings for DeepSeek V4.1 Flash 1017
{
"provider": {
"order": [
"sail-research/fp4",
"primeintellect",
"ionstream"
],
"ignore": [
"together",
"novita/fp8",
"atlas-cloud/fp8",
"gmicloud/fp8"
]
}
}Point your agent at llms.txt
Get updates when its successor drops
And which provider to use.
For model makers
Broken tool calls carry your model's name
Releasing a model? Test it before launch.
as often on GMICloud (fp8) as on Sail Research (fp4), its best provider.
broken tool calls a day across its 32 providers.