LLM gateways and observability
Asked as “LLM gateway”, and as “LLM observability and routing platform”, on behalf of a mid-market B2B company. 53 first choices recorded across the direct, paraphrase, budget and scale prompts, fourteen models each.
Standing · first-choice share
36%
Contested · Portkey 32%
36LiteLLM32Portkey06Bifrost26others
36% of first choices, contested.
Since October 2026↗new leaderNew leader since October 2026: LiteLLM (36%) replaces Portkey (33% then, 32% now), 4 points clear, inside the 11-point floor.LiteLLM leads at 36%, replacing Portkey, which led at 33% and stands at 32% now: 4 points clear, inside the floor, so the swap reads as unsettled.
By buyer segment
The same question asked on behalf of a different buyer. Each standing is computed within its segment; they sit side by side and are never added together.
Small business
Portkey36%Contested↗new leaderNew leader since October 2026: Portkey (36%) replaces LiteLLM (31% then, 33% now), 3 points clear, inside the 11-point floor.Standing →Mid-market
LiteLLM36%Contested↗new leaderNew leader since October 2026: LiteLLM (36%) replaces Portkey (33% then, 32% now), 4 points clear, inside the 11-point floor.This pageEnterprise
Bifrost23%Contested↗new leaderNew leader since October 2026: Bifrost (23%) replaces Portkey (26% then, 14% now), 9 points clear, inside the 11-point floor.Standing →
Portkey36%Contested↗new leaderNew leader since October 2026: Portkey (36%) replaces LiteLLM (31% then, 33% now), 3 points clear, inside the 11-point floor.Standing →Mid-marketLiteLLM36%Contested↗new leaderNew leader since October 2026: LiteLLM (36%) replaces Portkey (33% then, 32% now), 4 points clear, inside the 11-point floor.This pageEnterpriseBifrost23%Contested↗new leaderNew leader since October 2026: Bifrost (23%) replaces Portkey (26% then, 14% now), 9 points clear, inside the 11-point floor.Standing →The standing
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. Ordered by share.
ProductFirst-choice shareNegative rateLabelsQuadrantSince October 2026
01LiteLLM36%14%66endorsed leader▲+5Since October 2026: 31% → 36%, +5 points. Inside the 11-point floor: within noise.31% → 36%
LiteLLM36%14%66endorsed leader▲+5Since October 2026: 31% → 36%, +5 points. Inside the 11-point floor: within noise.31% → 36%02Portkey32%9%53endorsed leader▼−1Since October 2026: 33% → 32%, −1 point. Inside the 11-point floor: within noise.33% → 32%
Portkey32%9%53endorsed leader▼−1Since October 2026: 33% → 32%, −1 point. Inside the 11-point floor: within noise.33% → 32%03Bifrost6%3%33accepted challenger▲+2Since October 2026: 4% → 6%, +2 points. Inside the 11-point floor: within noise.4% → 6%
Bifrost6%3%33accepted challenger▲+2Since October 2026: 4% → 6%, +2 points. Inside the 11-point floor: within noise.4% → 6%04
Helicone4%6%16accepted challenger▼−2Since October 2026: 6% → 4%, −2 points. Inside the 11-point floor: within noise.6% → 4%
Helicone4%6%16accepted challenger▼−2Since October 2026: 6% → 4%, −2 points. Inside the 11-point floor: within noise.6% → 4%05
Cloudflare AI Gateway2%0%33accepted challenger=heldSince October 2026: 2% → 2%, ±0 points. Inside the 11-point floor: within noise.2% → 2%
Cloudflare AI Gateway2%0%33accepted challenger=heldSince October 2026: 2% → 2%, ±0 points. Inside the 11-point floor: within noise.2% → 2%06
OpenRouter2%20%41accepted challenger▼−7Since October 2026: 9% → 2%, −7 points. Inside the 11-point floor: within noise.9% → 2%
OpenRouter2%20%41accepted challenger▼−7Since October 2026: 9% → 2%, −7 points. Inside the 11-point floor: within noise.9% → 2%07
Langfuse2%0%16accepted challenger▲+2Since October 2026: 0% → 2%, +2 points. Inside the 11-point floor: within noise.0% → 2%
Langfuse2%0%16accepted challenger▲+2Since October 2026: 0% → 2%, +2 points. Inside the 11-point floor: within noise.0% → 2%Show the two products at 0%, ordered by negative rate
09
Vercel AI Gateway0%8%13accepted challenger=heldSince October 2026: 0% → 0%, ±0 points. Inside the 11-point floor: within noise.0% → 0%
Vercel AI Gateway0%8%13accepted challenger=heldSince October 2026: 0% → 0%, ±0 points. Inside the 11-point floor: within noise.0% → 0%08
Kong AI Gateway0%3%29accepted challenger=heldSince October 2026: 0% → 0%, ±0 points. Inside the 11-point floor: within noise.0% → 0%
Kong AI Gateway0%3%29accepted challenger=heldSince October 2026: 0% → 0%, ±0 points. Inside the 11-point floor: within noise.0% → 0%The floor is 11 points of share, measured: how far the models move a leader on their own when the same questions are asked twice with nothing changed. A larger change is movement; a smaller one can't be told apart from repeat variation, and both are shown. Each index measures its own floor from its own repeat, over its own categories, so the floor differs from one index to the next. Both editions asked the same fourteen models. How the floor is measured
Bars are the share of first choices, 0 to 100Every product with at least 10 labels here. Every product name links to its product page.
All twenty-one head-to-head pages: the top seven products, each against each
LiteLLM vs Cloudflare AI GatewayLiteLLM vs OpenRouterLiteLLM vs LangfusePortkey vs BifrostPortkey vs HeliconePortkey vs Cloudflare AI GatewayPortkey vs OpenRouterPortkey vs LangfuseBifrost vs HeliconeBifrost vs Cloudflare AI GatewayBifrost vs OpenRouterBifrost vs LangfuseHelicone vs Cloudflare AI GatewayHelicone vs OpenRouterHelicone vs LangfuseCloudflare AI Gateway vs OpenRouterCloudflare AI Gateway vs LangfuseOpenRouter vs Langfuse
Recommended versus criticized
Every product with at least 10 labels here, on both axes. The 30% line names a quadrant, not the verdict above: that one needs more than 40%.
Criticized challengerCriticized default
Negative label rate →
010203040506070809Accepted challengerEndorsed leader
0%First-choice share → · lines at 30% share and 25% negative50%
Key
01LiteLLM36%
LiteLLM36%02Portkey32%
Portkey32%03Bifrost6%
Bifrost6%04Helicone4%
Helicone4%05Cloudflare AI Gateway2%
Cloudflare AI Gateway2%06OpenRouter2%
OpenRouter2%07Langfuse2%
Langfuse2%08Kong AI Gateway0%
Kong AI Gateway0%09Vercel AI Gateway0%
Vercel AI Gateway0%What they warned about
Four of fourteen models held their first choice under the paraphrase. Claude Haiku 4.5, GPT-5.4 mini, Gemini 3.5 Flash, DeepSeek V4 Flash, Llama 4 Maverick, Qwen 3.7 Flash, Kimi K2, GLM 4.7 FlashX, MiniMax M2.5 and Muse Glimmer 30B changed. A high negative share on a product with few labels is a warning. A low share on a product with many labels is salience, not sentiment.
LiteLLM14%
9 of 66 labels negative · 9 of 14 models · 5 hard negative
“"LiteLLM put a credential stealer on PyPI and Docker on the day of my test, its second compromised release this year." [6]” Muse Glimmer 30B, negative prompt
OpenRouter20%
8 of 41 labels negative · 7 of 14 models · 2 hard negative
“The one I would not recommend for a mid-market B2B company is OpenRouter” DeepSeek V4 Flash, direct prompt
Portkey9%
5 of 53 labels negative · 4 of 14 models
“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
Azure API Management11%
1 of 9 labels negative · 1 of 14 models · 1 hard negative
“Avoid Overkill: Skip pure enterprise tools like Azure APIM unless you're Microsoft-heavy.” Grok 4.20, direct prompt
What they cite
Fourteen of the fourteen models returned a source list in this edition: five return the pages they cited and nine the search results they consulted.
Sites the answers cite
74 of 84 answers in this category came back with a source list, from 14 of 14 models: citations where the model returns them, or the search results it consulted. 869 links across 210 sites, every framing counted. Ranked by the number of answers carrying the site or page. 38 of the 252 answers across every segment cited this index's own page for the category; the method page measures whether that reading tilts an answer.
46 answers · 126 citations · 12 models
39 answers · 71 citations · 12 models
vendor site · GitHub23 answers · 59 citations · 8 models
vendor site · LLMGateway.io19 answers · 33 citations · 10 models
vendor site · Requesty15 answers · 16 citations · 10 models
vendor site · Respan15 answers · 16 citations · 10 models
vendor site · OpenRouter13 answers · 19 citations · 6 models
vendor site · Braintrust11 answers · 19 citations · 9 models
vendor site · LiteLLM11 answers · 15 citations · 7 models
publisher (Gane)10 answers · 12 citations · 6 models
vendor site · Arize10 answers · 10 citations · 7 models
vendor site · CloudZero10 answers · 10 citations · 5 models
Pages the answers cite
The ten pages named in the most answers, by full address. A page here is one the models returned with a recommendation, not one the index endorses.
11 answers · 8 models
10 answers · 5 models
9 answers · 6 models
9 answers · 6 models
9 answers · 6 models
9 answers · 6 models
9 answers · 6 models
8 answers · 6 models
7 answers · 4 models
7 answers · 6 models
Search against answers
Each company's standing in the answers beside its site's footprint in Google search, one row a site: the products the models named on it with their shares, and the share they add up to; monthly searches on Google, and DataForSEO's estimate of AI search demand (modeled from search signals, directional, not a count of queries to any assistant), for the most-searched of the company's and its products' names (the name is in each row's hover text); estimated monthly organic visits to the site; and its best position in Google's top ten for “best llm gateway”, “llm gateway”, “llm gateways”. US estimates from DataForSEO and Google's Ads Transparency Center. A small company's site, or a mid-sized company's site for its flagship, is marked company; a product on a large parent's site (Google, Microsoft) has no site figures. A column with no figures for this category is left out, and an empty cell means none were seen, not none exist. Two measurements side by side: neither is read as the cause of the other.
| Company and products | Share | Own site cited | Name searches, Google | AI search demand, est. | Organic visits | Google ads, last 30 days | Best Google position |
|---|---|---|---|---|---|---|---|
| LiteLLM litellm.ai | 36% | 196 | 22,200 | 264 | 15,281 | 0 | 2 |
| Portkey portkey.ai | 32% | 156 | 5,400 | 131 | 5,999 | 0 | 5 |
| Bifrost getmaxim.ai | 6% | 120 | 12,100 | 427 | 4,289 | 10 | 8 |
| Helicone helicone.ai | 4% | 50 | 1,600 | 899 | 2,439 | 0 | |
| Cloudflare parent site Cloudflare AI Gateway 2% | 2% | 96 | 550,000 | 10,897 | |||
| OpenRouter openrouter.ai | 2% | 118 | 135,000 | 655 | 225,878 | 68 | 7 |
| Langfuse langfuse.com | 2% | 48 | 18,100 | 216 | 8,530 | 4 | |
| LangSmith langchain.com | 2% | 19 | 14,800 | 238 | 93,424 | 94 | |
| Braintrust Gateway braintrust.dev | 2% | 15 | 30 | 0 | 14,321 | 69 | 5 |
| Respan respan.ai | 2% | 11 |
Google's top ten for the three searches hold 18 sites; 6 of them are among the sites the models cited here (braintrust.dev, dev.to, getmaxim.ai, github.com, llmgateway.io, openrouter.ai). The first result for “best llm gateway” is reddit.com.
How they answered
Six framings of the same buying question, each sent to every model in a fresh session with search on. One row per model, so a row shows whether it held its answer under rewording, what it named when cost was the constraint, and what it argued against. Computed from the raw judge labels.ShowHide
How they answered
Six framings of the same buying question, each sent to every model in a fresh session with search on. One row per model, so a row shows whether it held its answer under rewording, what it named when cost was the constraint, and what it argued against. Computed from the raw judge labels.
ShowHideModelDirect“What is the best LLM gateway for a mid-market B2B company?”Paraphrase“Which LLM observability and routing platform would you recommend to a mid-sized B2B company?”Comparative“What are the top LLM gateways and how do they differ?”Budget-constrained“What is the best LLM gateway for a company with a limited budget?”Scale-constrained“We are a 500 person company evaluating a LLM gateway. What should we look at?”Negative“Which LLM gateways should I avoid or be cautious about?”
Claude Haiku 4.5Nexos.ai, Portkeyno first choiceChangedBifrostLocalAI, Ollamano first choicenothing named
Three alternatives
Bifrost, LiteLLM, OpenRouterFive alternatives
Cloudflare AI Gateway, Kong AI Gateway, LiteLLM, OpenRouter, PortkeyOne alternative
OpenRouterGPT-5.4 miniOpenRouterLangSmith, LiteLLMChangedLiteLLMLiteLLMagainst: Kong AI Gateway, Portkey, TrueFoundry AI Gatewayno first choicenothing named
Three alternatives
Kong AI Gateway, LiteLLM, PortkeyThree alternatives
Datadog LLM Observability, Langfuse, OpenRouterFour alternatives
Kong AI Gateway, OpenRouter, Portkey, Vercel AI GatewayTwo alternatives
Cloudflare AI Gateway, OpenRouterGemini 3.5 FlashLiteLLM, PortkeyPortkeyChangedLiteLLMLiteLLMLiteLLM, Portkeyagainst: LiteLLM
Two alternatives
Bifrost, Cloudflare AI GatewayThree alternatives
Braintrust Gateway, Langfuse, LiteLLMSeven alternatives
Apache APISIX, Bifrost, Braintrust Gateway, Cloudflare AI Gateway, Kong AI Gateway, OpenRouter, PortkeyTwo alternatives
Bifrost, Cloudflare AI GatewayFour alternatives
Arch Gateway, Cloudflare AI Gateway, Helicone, Kong AI GatewayPerplexity SonarPortkeyPortkeyHeldLiteLLMagainst: RouteLLMLiteLLMno first choiceagainst: LiteLLM, OpenRouter, Portkey, Vercel AI Gateway
Three alternatives
Bifrost, LiteLLM, OpenRouterFour alternatives
Bifrost, Braintrust Gateway, LangSmith, LangfuseSix alternatives
Bifrost, Cloudflare AI Gateway, Kong AI Gateway, OpenRouter, Portkey, Vercel AI GatewayFour alternatives
Cloudflare AI Gateway, OpenRouter, Portkey, Vercel AI GatewayGrok 4.20Portkeyagainst: Azure API Management, BifrostPortkeyHeldBifrost, LiteLLM, PortkeyLiteLLMno first choiceagainst: Helicone, LiteLLM, PromptLayer
Two alternatives
Cloudflare AI Gateway, LiteLLMFour alternatives
Helicone, LangSmith, Langfuse, LiteLLMFour alternatives
Cloudflare AI Gateway, Kong AI Gateway, OpenRouter, Vercel AI GatewayThree alternatives
Bifrost, Helicone, LLM GatewayMistral SmallLiteLLM, PortkeyLiteLLM, PortkeyHeldno first choiceStandard Computeno first choiceagainst: LiteLLM, Ollama
Two alternatives
OpenRouter, RequestyTwo alternatives
Braintrust Gateway, RespanThree alternatives
Bifrost, Future AGI Agent Command Center, LiteLLMDeepSeek V4 FlashBifrost, Portkeyagainst: OpenRouterOrq.aiChangedBifrostLiteLLMno first choiceagainst: LiteLLM
Three alternatives
Amazon Bedrock, Azure API Management, LiteLLMFour alternatives
LangSmith, Langfuse, LiteLLM, PortkeySix alternatives
Cloudflare AI Gateway, Kong AI Gateway, LiteLLM, OpenRouter, Portkey, Vercel AI GatewayTwo alternatives
Cloudflare AI Gateway, OpenRouterLlama 4 MaverickFastRouterno first choiceChangedBifrostno first choiceno first choicenothing named
Four alternatives
Concentrate, LiteLLM, OpenRouter, PortkeyFour alternatives
Cloudflare AI Gateway, Kong AI Gateway, LiteLLM, OpenRouterQwen 3.7 FlashLiteLLM, PortkeyHeliconeChangedLiteLLMagainst: OpenRouterLiteLLMagainst: Merge.dev, PortkeyPortkeyagainst: LiteLLM, OpenRouter, Portkey
Four alternatives
Bifrost, Cloudflare AI Gateway, Kong AI Gateway, OpenRouterTwo alternatives
Braintrust Gateway, LangfuseFour alternatives
Bifrost, Cloudflare AI Gateway, Kong AI Gateway, Vercel AI GatewayTwo alternatives
Cloudflare AI Gateway, OpenRouterTwo alternatives
Latenode, Voyage AIKimi K2PortkeyHeliconeChangedBifrostLiteLLMagainst: OpenRouterno first choiceagainst: AIgateway, LiteLLM, OpenGateway
Four alternatives
FastRouter, Kong AI Gateway, LiteLLM, TrueFoundry AI GatewayOne alternative
PortkeySeven alternatives
Cloudflare AI Gateway, Envoy AI Gateway, Kong AI Gateway, LiteLLM, Portkey, Tetrate Agent Router Enterprise, TrueFoundry AI GatewayTwo alternatives
Cloudflare AI Gateway, Vercel AI GatewayGLM 4.7 FlashXLiteLLMBifrost, LangfuseChangedBifrostagainst: OpenRouterBifrost, LiteLLMno first choicenothing named
Three alternatives
Bifrost, Kong AI Gateway, PortkeySix alternatives
Arize Phoenix, Confident AI, Kong AI Gateway, LangSmith, LiteLLM, PortkeyThirteen alternatives
AI Gateway by GitLab, Cloudflare AI Gateway, Gloo Gateway, IBM API Connect AI Gateway, Kong AI Gateway, LM Studio, LiteLLM, Portkey, TrueFoundry AI Gateway, Tyk, Vercel AI Gateway, Wealthsimple LLM Gateway, aigateway.appThree alternatives
Cloudflare AI Gateway, OpenRouter, Vercel AI GatewayMiniMax M2.5LiteLLMBraintrust GatewayChangedBifrostLiteLLMagainst: Portkeyno first choicenothing named
Four alternatives
Cloudflare AI Gateway, Helicone, OpenRouter, PortkeyTwo alternatives
LangSmith, LangfuseFive alternatives
Cloudflare AI Gateway, Kong AI Gateway, LiteLLM, OpenRouter, ZuploTwo alternatives
Cloudflare AI Gateway, LLM GatewayGPT-6 LunaPortkeyPortkeyHeldno first choiceCloudflare AI Gatewayno first choiceagainst: LiteLLM, OpenRouter
Two alternatives
Cloudflare AI Gateway, LiteLLMTwo alternatives
Langfuse, LiteLLMSix alternatives
Cloudflare AI Gateway, Kong AI Gateway, LiteLLM, OpenRouter, Portkey, Vercel AI GatewayTwo alternatives
LiteLLM OSS, OpenRouterMuse Glimmer 30BLiteLLM, PortkeyRespanChangedBifrostLiteLLMno first choiceagainst: LiteLLM, OpenRouter
Two alternatives
Kong AI Gateway, TrueFoundry AI GatewayTwo alternatives
Langfuse, PortkeySeven alternatives
Cloudflare AI Gateway, Kong AI Gateway, LiteLLM, OpenRouter, Portkey, TrueFoundry AI Gateway, Vercel AI GatewayFive alternatives
Cloudflare AI Gateway, LLM Gateway, LLMGateway.io, OpenRouter, VercelBold is the first choiceAlternatives are counted; the count opens them.What the answer argued against
The record
One row per call: the version string exactly as returned, whether the model searched, sources cited, and latency.
Eighty-four rows: every prompt, every model, every answer.
| Prompt | Model | Version string | Time (UTC) | Searched | Sources | Latency |
|---|---|---|---|---|---|---|
| Direct recommendation | Claude Haiku 4.5 | claude-haiku-4-5-20251001 | 2026-11-01 10:46 | yes | 19 | 9 s |
| Direct recommendation | GPT-5.4 mini | gpt-5.4-mini-2026-03-17 | 2026-11-01 16:28 | yes | 3 | 5 s |
| Direct recommendation | Gemini 3.5 Flash | gemini-3.6-flash (asked for gemini-3.5-flash; the API answered as this version) | 2026-11-01 11:27 | yes | 9 | 27 s |
| Direct recommendation | Perplexity Sonar | sonar | 2026-11-01 15:48 | yes | 10 | 4 s |
| Direct recommendation | Grok 4.20 | spacexai/grok-4.20-non-reasoning via xai | 2026-11-01 14:43 | yes | 13 | 18 s |
| Direct recommendation | Mistral Small | mistral/mistral-small via mistral | 2026-11-01 17:53 | yes | 5 | 4 s |
| Direct recommendation | DeepSeek V4 Flash | deepseek/deepseek-v4-flash via deepinfra | 2026-11-01 16:59 | yes | 24 | 34 s |
| Direct recommendation | Llama 4 Maverick | meta/llama-4-maverick via bedrock | 2026-11-01 09:19 | yes | 5 | 3 s |
| Direct recommendation | Qwen 3.7 Flash | alibaba/qwen3.7-flash via alibaba | 2026-11-01 09:10 | yes | 5 | 35 s |
| Direct recommendation | Kimi K2 | moonshotai/kimi-k2 via novita | 2026-11-01 11:51 | yes | 15 | 43 s |
| Direct recommendation | GLM 4.7 FlashX | zai/glm-4.7-flashx via zai | 2026-11-01 13:56 | yes | 23 | 46 s |
| Direct recommendation | MiniMax M2.5 | minimax/minimax-m2.5 via minimax | 2026-11-01 09:43 | yes | 5 | 32 s |
| Direct recommendation | GPT-6 Luna | gpt-6-luna | 2026-11-01 08:54 | yes | 3 | 13 s |
| Direct recommendation | Muse Glimmer 30B | meta/muse-glimmer-30b via togetherai | 2026-11-01 14:22 | yes | 12 | 15 s |
| Paraphrase | Claude Haiku 4.5 | claude-haiku-4-5-20251001 | 2026-11-01 17:33 | no | 0 | 3 s |
| Paraphrase | GPT-5.4 mini | gpt-5.4-mini-2026-03-17 | 2026-11-01 15:13 | yes | 5 | 5 s |
| Paraphrase | Gemini 3.5 Flash | gemini-3.6-flash (asked for gemini-3.5-flash; the API answered as this version) | 2026-11-01 14:54 | no | 0 | 18 s |
| Paraphrase | Perplexity Sonar | sonar | 2026-11-01 17:12 | yes | 10 | 4 s |
| Paraphrase | Grok 4.20 | spacexai/grok-4.20-non-reasoning via xai | 2026-11-01 17:26 | yes | 10 | 10 s |
| Paraphrase | Mistral Small | mistral/mistral-small via mistral | 2026-11-01 13:09 | yes | 5 | 4 s |
| Paraphrase | DeepSeek V4 Flash | deepseek/deepseek-v4-flash via deepinfra | 2026-11-01 13:00 | yes | 24 | 27 s |
| Paraphrase | Llama 4 Maverick | meta/llama-4-maverick via bedrock | 2026-11-01 15:24 | yes | 5 | 3 s |
| Paraphrase | Qwen 3.7 Flash | alibaba/qwen3.7-flash via alibaba | 2026-11-01 12:24 | yes | 10 | 22 s |
| Paraphrase | Kimi K2 | moonshotai/kimi-k2 via novita | 2026-11-01 18:45 | yes | 25 | 27 s |
| Paraphrase | GLM 4.7 FlashX | zai/glm-4.7-flashx via zai | 2026-11-01 17:40 | yes | 20 | 24 s |
| Paraphrase | MiniMax M2.5 | minimax/minimax-m2.5 via minimax | 2026-11-01 13:59 | yes | 23 | 97 s |
| Paraphrase | GPT-6 Luna | gpt-6-luna | 2026-11-01 15:54 | yes | 3 | 16 s |
| Paraphrase | Muse Glimmer 30B | meta/muse-glimmer-30b via togetherai | 2026-11-01 18:10 | yes | 15 | 24 s |
| Comparative | Claude Haiku 4.5 | claude-haiku-4-5-20251001 | 2026-11-01 11:57 | yes | 18 | 9 s |
| Comparative | GPT-5.4 mini | gpt-5.4-mini-2026-03-17 | 2026-11-01 18:55 | yes | 6 | 8 s |
| Comparative | Gemini 3.5 Flash | gemini-3.6-flash (asked for gemini-3.5-flash; the API answered as this version) | 2026-11-01 19:16 | yes | 11 | 25 s |
| Comparative | Perplexity Sonar | sonar | 2026-11-01 10:10 | yes | 10 | 5 s |
| Comparative | Grok 4.20 | spacexai/grok-4.20-non-reasoning via xai | 2026-11-01 16:51 | yes | 5 | 11 s |
| Comparative | Mistral Small | mistral/mistral-small via mistral | 2026-11-01 12:42 | yes | 5 | 8 s |
| Comparative | DeepSeek V4 Flash | deepseek/deepseek-v4-flash via deepinfra | 2026-11-01 17:54 | yes | 25 | 30 s |
| Comparative | Llama 4 Maverick | meta/llama-4-maverick via bedrock | 2026-11-01 16:15 | yes | 5 | 2 s |
| Comparative | Qwen 3.7 Flash | alibaba/qwen3.7-flash via alibaba | 2026-11-01 09:19 | yes | 10 | 25 s |
| Comparative | Kimi K2 | moonshotai/kimi-k2 via novita | 2026-11-01 08:51 | yes | 15 | 59 s |
| Comparative | GLM 4.7 FlashX | zai/glm-4.7-flashx via zai | 2026-11-01 13:02 | yes | 25 | 120 s |
| Comparative | MiniMax M2.5 | minimax/minimax-m2.5 via minimax | 2026-11-01 11:13 | yes | 10 | 16 s |
| Comparative | GPT-6 Luna | gpt-6-luna | 2026-11-01 10:22 | yes | 6 | 20 s |
| Comparative | Muse Glimmer 30B | meta/muse-glimmer-30b via togetherai | 2026-11-01 13:30 | yes | 15 | 37 s |
| Budget constrained | Claude Haiku 4.5 | claude-haiku-4-5-20251001 | 2026-11-01 15:51 | no | 0 | 5 s |
| Budget constrained | GPT-5.4 mini | gpt-5.4-mini-2026-03-17 | 2026-11-01 10:51 | yes | 5 | 5 s |
| Budget constrained | Gemini 3.5 Flash | gemini-3.6-flash (asked for gemini-3.5-flash; the API answered as this version) | 2026-11-01 19:40 | yes | 11 | 22 s |
| Budget constrained | Perplexity Sonar | sonar | 2026-11-01 11:46 | yes | 10 | 4 s |
| Budget constrained | Grok 4.20 | spacexai/grok-4.20-non-reasoning via xai | 2026-11-01 15:37 | yes | 5 | 7 s |
| Budget constrained | Mistral Small | mistral/mistral-small via mistral | 2026-11-01 10:03 | yes | 5 | 5 s |
| Budget constrained | DeepSeek V4 Flash | deepseek/deepseek-v4-flash via deepinfra | 2026-11-01 12:12 | yes | 25 | 27 s |
| Budget constrained | Llama 4 Maverick | meta/llama-4-maverick via bedrock | 2026-11-01 17:32 | yes | 7 | 4 s |
| Budget constrained | Qwen 3.7 Flash | alibaba/qwen3.7-flash via alibaba | 2026-11-01 17:17 | yes | 5 | 43 s |
| Budget constrained | Kimi K2 | moonshotai/kimi-k2 via novita | 2026-11-01 19:37 | yes | 10 | 18 s |
| Budget constrained | GLM 4.7 FlashX | zai/glm-4.7-flashx via zai | 2026-11-01 15:22 | yes | 23 | 76 s |
| Budget constrained | MiniMax M2.5 | minimax/minimax-m2.5 via minimax | 2026-11-01 13:38 | yes | 5 | 30 s |
| Budget constrained | GPT-6 Luna | gpt-6-luna | 2026-11-01 19:15 | yes | 4 | 19 s |
| Budget constrained | Muse Glimmer 30B | meta/muse-glimmer-30b via togetherai | 2026-11-01 13:08 | yes | 20 | 28 s |
| Scale constrained | Claude Haiku 4.5 | claude-haiku-4-5-20251001 | 2026-11-01 13:56 | no | 0 | 6 s |
| Scale constrained | GPT-5.4 mini | gpt-5.4-mini-2026-03-17 | 2026-11-01 17:53 | no | 0 | 7 s |
| Scale constrained | Gemini 3.5 Flash | gemini-3.6-flash (asked for gemini-3.5-flash; the API answered as this version) | 2026-11-01 11:19 | no | 0 | 23 s |
| Scale constrained | Perplexity Sonar | sonar | 2026-11-01 13:05 | yes | 10 | 6 s |
| Scale constrained | Grok 4.20 | spacexai/grok-4.20-non-reasoning via xai | 2026-11-01 14:04 | yes | 5 | 12 s |
| Scale constrained | Mistral Small | mistral/mistral-small via mistral | 2026-11-01 08:41 | no | 0 | 7 s |
| Scale constrained | DeepSeek V4 Flash | deepseek/deepseek-v4-flash via deepinfra | 2026-11-01 19:29 | yes | 25 | 48 s |
| Scale constrained | Llama 4 Maverick | meta/llama-4-maverick via bedrock | 2026-11-01 15:08 | yes | 5 | 3 s |
| Scale constrained | Qwen 3.7 Flash | alibaba/qwen3.7-flash via alibaba | 2026-11-01 13:40 | no | 0 | 33 s |
| Scale constrained | Kimi K2 | moonshotai/kimi-k2 via novita | 2026-11-01 12:11 | yes | 10 | 18 s |
| Scale constrained | GLM 4.7 FlashX | zai/glm-4.7-flashx via zai | 2026-11-01 13:32 | yes | 24 | 202 s |
| Scale constrained | MiniMax M2.5 | minimax/minimax-m2.5 via minimax | 2026-11-01 19:14 | yes | 10 | 26 s |
| Scale constrained | GPT-6 Luna | gpt-6-luna | 2026-11-01 14:53 | yes | 3 | 17 s |
| Scale constrained | Muse Glimmer 30B | meta/muse-glimmer-30b via togetherai | 2026-11-01 11:33 | yes | 15 | 25 s |
| Negative framing | Claude Haiku 4.5 | claude-haiku-4-5-20251001 | 2026-11-01 16:06 | no | 0 | 4 s |
| Negative framing | GPT-5.4 mini | gpt-5.4-mini-2026-03-17 | 2026-11-01 17:57 | yes | 7 | 7 s |
| Negative framing | Gemini 3.5 Flash | gemini-3.6-flash (asked for gemini-3.5-flash; the API answered as this version) | 2026-11-01 17:32 | yes | 12 | 20 s |
| Negative framing | Perplexity Sonar | sonar | 2026-11-01 10:32 | yes | 10 | 4 s |
| Negative framing | Grok 4.20 | spacexai/grok-4.20-non-reasoning via xai | 2026-11-01 14:31 | yes | 5 | 14 s |
| Negative framing | Mistral Small | mistral/mistral-small via mistral | 2026-11-01 10:44 | yes | 5 | 4 s |
| Negative framing | DeepSeek V4 Flash | deepseek/deepseek-v4-flash via deepinfra | 2026-11-01 08:51 | yes | 22 | 23 s |
| Negative framing | Llama 4 Maverick | meta/llama-4-maverick via bedrock | 2026-11-01 17:40 | yes | 5 | 3 s |
| Negative framing | Qwen 3.7 Flash | alibaba/qwen3.7-flash via alibaba | 2026-11-01 13:35 | yes | 14 | 30 s |
| Negative framing | Kimi K2 | moonshotai/kimi-k2 via novita | 2026-11-01 10:54 | yes | 25 | 50 s |
| Negative framing | GLM 4.7 FlashX | zai/glm-4.7-flashx via zai | 2026-11-01 15:40 | yes | 23 | 86 s |
| Negative framing | MiniMax M2.5 | minimax/minimax-m2.5 via minimax | 2026-11-01 13:14 | no | 0 | 13 s |
| Negative framing | GPT-6 Luna | gpt-6-luna | 2026-11-01 14:05 | yes | 2 | 21 s |
| Negative framing | Muse Glimmer 30B | meta/muse-glimmer-30b via togetherai | 2026-11-01 16:01 | yes | 25 | 34 s |
Normalization in this category
Every judgment call made between the raw labels and the numbers above, listed so it is visible and reversible.
ShowHide
Category-scoped readings
Braintrust read as Braintrust Gateway
Cloudflare read as Cloudflare AI Gateway
Kong read as Kong AI Gateway
Kong Enterprise read as Kong AI Gateway
LLM Gateway.io read as LLM Gateway
Portkey.ai read as Portkey
TrueFoundry read as TrueFoundry AI Gateway
Unresolved, counted raw
AI Gateway by GitLab
AIgateway
Anthropic’s own gateway features
Arch Gateway
Azure AI API Management
Blackbox AI
Eden AI
F5 AI Gateway
GateScope
Google AI
Google Cloud API gateways
Google Cloud's Agent Platform
HuggingFace Inference API
IBM API Connect AI Gateway
Kor
Latenode
LiteLLM OSS
Merge.dev
OpenGateway
Perplexity API
Portkey OSS
Ray Serve
RedPill
RouteLLM
Standard Compute
Vellum AI
Voyage AI
Wealthsimple LLM Gateway
aigateway.app
Discontinued, still offered
No shut-down product was recommended here.