Ten of fourteen models named Portkey first on the direct prompt; one named Bifrost. Portkey was named by fourteen of the fourteen models and Bifrost by twelve and Portkey carries 53 labels and Bifrost 33, so the shares are not directly comparable.
Named in two categories this edition.
Named in four categories this edition.
Share is the count of first choices across the direct, paraphrase, budget and scale prompts over all fourteen models, for a mid-market B2B company; rank is within the category; every quote names the model and the prompt it came from. Both figures come from the LLM gateways and observability page.
Across every category in the November 2026 Edition, Portkey and Bifrost were named in the same answer seventy-four times, of the 169 answers naming Portkey and the 121 naming Bifrost. In those answers Bifrost took the first choice twenty-four times and Portkey seventeen.
| Model | Direct | Paraphrase | Comparative | Budget-constrained | Scale-constrained | Negative |
|---|---|---|---|---|---|---|
| Claude Haiku 4.5 | ||||||
| GPT-5.4 mini | ||||||
| Gemini 3.5 Flash | ||||||
| Perplexity Sonar | ||||||
| Grok 4.20 | ||||||
| Mistral Small | ||||||
| DeepSeek V4 Flash | ||||||
| Llama 4 Maverick | ||||||
| Qwen 3.7 Flash | ||||||
| Kimi K2 | ||||||
| GLM 4.7 FlashX | ||||||
| MiniMax M2.5 | ||||||
| GPT-6 Luna | ||||||
| Muse Glimmer 30B |
Bold names in an answer are the products the judge labeled a first choice; a model naming several gives each of them that label. The full answer text for every row is in the record.
Every negative label with a quote, up to three, then the highest-weighted positives, up to three. Six of eight in this category shown.
“Avoid percentage-fee gateways until your monthly model spend exceeds $10,000–$35,000, as the fees become cost-prohibitive” MiniMax M2.5 · budget prompt · soft negative
“Providers like standard versions of OpenRouter or Portkey typically process logs across public infrastructure.” Qwen 3.7 Flash · negative prompt · soft negative
“It eliminates the subscription cost associated with commercial gateways (like Merge.dev or Portkey).” Qwen 3.7 Flash · budget prompt · soft negative
“the "best" LLM gateway typically comes down to a choice between the open-source standard, LiteLLM, or the observability-first solution, Portkey” Qwen 3.7 Flash · direct prompt · first choice
“consider starting with self-hosted open-source options (like Portkey's MIT-licensed tier or Envoy AI Gateway) where you control the data plane” Kimi K2 · negative prompt · first choice
“Portkey has long been the mid-market pick for teams that want guardrails, caching and logging in one product” Claude Haiku 4.5 · direct prompt · first choice
Every negative label with a quote, up to three, then the highest-weighted positives, up to three. Four of five in this category shown.
“it may be overkill (and more enterprise-focused) for simpler mid-market needs” Grok 4.20 · direct prompt · soft negative
“Bifrost leads on published performance (11 microseconds of overhead at 5,000 RPS) and is the only one here with governance and an MCP gateway in the same open-source binary.” Muse Glimmer 30B · comparative prompt · first choice
“Bifrost leads for production at a benchmarked 11 microseconds of overhead, with a Go core, native MCP gateway, semantic caching, and virtual-key governance” Claude Haiku 4.5 · comparative prompt · first choice
“Bifrost leads in performance with 11 microseconds of overhead at 5,000 RPS and offers advanced governance features.” Llama 4 Maverick · comparative prompt · first choice
Comparisons are drawn for the top eight products in each category, each against each. The output is the models' output; nothing here is a recommendation by the index.