AI models

Currently this platform charges no fees

Category [optional]: AI models Cluster [optional]: United States Who: xAI. Grok models (Grok 4.20 Beta, Grok 4.5 / 4.1 Thinking, multi-agent betas in late 2026 arenas) distributed via x.com / grok product and API where offered. Strengths: real-time / social-context flavor, reasoning variants, multi-agent betas for tool workflows. Limits: brand and policy posture differ from enterprise-first vendors; verify commercial terms and data use for B2B. Desk tips: pin exact Grok build string; test toxicity / brand-safe filters on your prompts before catalog copy automation. Grok: https://grok.x.ai/ xAI: https://x.ai/

Anonymous · 2026-10-31 22:27

Submit a complaint

Currently this platform charges no fees

Category [optional]: AI models Cluster [optional]: China / Shenzhen Who: Tencent. Hunyuan family across WeChat/QQ ecosystem and Tencent Cloud; Hy3 open-marked rows (Apache 2.0) appear on LMArena mid ranks. Strengths: social-app distribution, cloud billing in CN, open Hy3 options for self-host experiments. Limits: product naming across WeChat plugin vs Cloud API can confuse POs — pin ids. Tencent Cloud Hunyuan: https://cloud.tencent.com/product/hunyuan

Anonymous · 2026-10-31 22:27

Submit a complaint

Currently this platform charges no fees

Category [optional]: AI models Cluster [optional]: China / Hangzhou Who: Alibaba Cloud Tongyi / Qwen. Closed Max/Plus APIs (Qwen3.8-Max ~top-25 LMArena Oct 2026) plus many Apache-2.0 open sizes for self-host (e.g. Qwen3.x dense/MoE releases). Strengths: Chinese + multilingual, long context options, cloud + open dual track, ecosystem on Model Studio / DashScope. Limits: license differs Max API vs open weights; enterprise features on Alibaba Cloud regions. Desk tips: name DashScope model id in PO; for OEM boxes ship the open weight checksum. Qwen: https://qwenlm.github.io/ Model Studio: https://www.alibabacloud.com/product/model-studio

Anonymous · 2026-10-31 22:27

Submit a complaint

Currently this platform charges no fees

Category [optional]: AI models / search Cluster [optional]: United States Who: Perplexity. Search-style answer engine with citations; Pro routes across multiple frontier models. Strengths: research desks, source links, lower hallucination risk vs bare chat when citations are checked. Limits: not a self-host LLM; API/terms for commercial republishing must be read carefully. Site: https://www.perplexity.ai/

Anonymous · 2026-10-31 22:27

Submit a complaint

Currently this platform charges no fees

Category [optional]: AI models Cluster [optional]: United States Who: OpenAI (US). Consumer product ChatGPT; developer surface API + Assistants/Agents; video Sora; image generation inside ChatGPT/API. What it is good at: general chat, coding assist, multimodal (text/image/voice), agents with tools, ecosystem of plugins and enterprise SSO. Late-2026 arena names include GPT-5.5 / GPT-5.6 Sol / GPT-6 Sol / Astra / Luna variants — always quote the exact model string in contracts. Limits: closed weights for flagship chat models; rate limits and region availability vary; content policy is strict; price tiers from Plus to Pro / API metered. Trade desk tips: ask for data residency, zero-retention options, SOC2/ISO docs, and whether fine-tunes leave their cloud. Pair coding RFQs with Cursor/Copilot columns. Site: https://openai.com/ API: https://platform.openai.com/ ChatGPT: https://chatgpt.com/

Anonymous · 2026-10-31 22:27

Submit a complaint

Currently this platform charges no fees

Category [optional]: AI models Cluster [optional]: European Union / France Who: Mistral AI (France/EU). Mix of open weights and commercial Large/Magistral APIs — preferred when EU data residency and open options matter in RFPs. Strengths: EU vendor story, Apache/open releases historically, Le Chat product, competitive coding. Limits: flagship closed APIs still metered; verify current open vs closed matrix per release. Site: https://mistral.ai/ La Plateforme: https://console.mistral.ai/

Anonymous · 2026-10-31 22:27

Submit a complaint

Currently this platform charges no fees

Category [optional]: AI models Cluster [optional]: China / Shanghai Who: MiniMax. Text models (MiniMax-M3 on arenas), plus strong speech and video product lines used in CN creator tools. Strengths: multimodal product suite, community license variants, creator-market distribution. Limits: read MiniMax Community License before OEM redistribution. Site: https://www.minimaxi.com/

Anonymous · 2026-10-31 22:27

Submit a complaint

Currently this platform charges no fees

Category [optional]: AI models Cluster [optional]: China Who: Xiaomi. MiMo-V2.6-Pro (MIT) and Flash variants show competitive Elo / price on Oct 2026 LMArena (~29 / mid ranks) — watch for edge/device narrative tying to Xiaomi hardware. Strengths: aggressive pricing, MIT on marked lines, potential on-device synergy. Limits: newer brand in frontier LLM RFPs — demand bake-off evidence. Desk tips: useful cost baseline vs DeepSeek/GLM in CN private cloud quotes. Check Xiaomi / MiMo release notes for the current download + API URLs when publishing an RFQ.

Anonymous · 2026-10-31 22:27

Submit a complaint

Currently this platform charges no fees

Category [optional]: AI models Cluster [optional]: United States / Global Microsoft cloud Who: Microsoft. Copilot across M365/Windows/GitHub; Azure OpenAI Service hosting OpenAI models in Microsoft tenancy; Phi small models for edge/on-prem experiments. Strengths: enterprise identity (Entra), compliance, GitHub Copilot for coding desks, familiar procurement. Limits: model quality follows upstream OpenAI/Phi releases; seat licensing complexity. Azure OpenAI: https://azure.microsoft.com/products/ai-services/openai-service Copilot: https://www.microsoft.com/microsoft-copilot GitHub Copilot: https://github.com/features/copilot

Anonymous · 2026-10-31 22:27

Submit a complaint

Currently this platform charges no fees

Category [optional]: AI models Cluster [optional]: United States Who: Meta AI. Late-2026 closed Muse Wonder (Max) and Muse Spark 1.x sit high on LMArena; Meta also remains central to open Llama-class weights that power many self-host stacks. Strengths: research-to-product scale, open-weights culture for Llama generations, multimodal experiments, WhatsApp/Instagram distribution for consumer AI. Limits: license terms differ by release — read each Llama / Muse license; API vs downloadable weights are not interchangeable. Desk tips: for factory on-prem, prefer documented open weights + your GPU vendor; for chat quality, quote Muse model id from the vendor API you actually buy. Meta AI: https://ai.meta.com/ Llama: https://www.llama.com/

Anonymous · 2026-10-31 22:27

Submit a complaint

Currently this platform charges no fees

Category [optional]: AI models Cluster [optional]: China / Beijing Who: Moonshot AI. Kimi consumer app + API; Kimi K3 (Max) ranked ~19 on LMArena Text Oct 2026; K2.x Thinking variants for longer reasoning. Strengths: long-context product narrative, Chinese product UX, competitive arena scores. Limits: check Kimi license labels (K3 license vs Modified MIT on older lines) before redistribution. Desk tips: good for CN-market chat desks; export contracts need counsel on data transfer. Kimi: https://kimi.moonshot.cn/ Moonshot: https://www.moonshot.cn/

Anonymous · 2026-10-31 22:27

Submit a complaint

Currently this platform charges no fees

Category [optional]: AI models Cluster [optional]: United States Who: Google DeepMind / Google AI. Gemini closed models (Gemini 4 Carbon/Argon High often top LMArena Text in Oct 2026), Flash cost/latency tiers, Gemma open-weights, Veo video with native audio, Imagen image line, Vertex AI for enterprise. Strengths: multimodal, long context (often 1M class), search/workspace integration, strong Flash price/performance, Vertex compliance story. Limits: some frontier variants marked preliminary on public arenas; enterprise features may need GCP contracting. Desk tips: separate consumer Gemini app vs Vertex model ids in RFQs; ask for region pinning and VPC-SC. Gemini: https://gemini.google.com/ AI Studio: https://aistudio.google.com/ Vertex AI: https://cloud.google.com/vertex-ai Gemma: https://ai.google.dev/gemma

Anonymous · 2026-10-31 22:27

Submit a complaint

Currently this platform charges no fees

Category [optional]: AI models Cluster [optional]: China / Beijing Who: Zhipu AI (Z.ai). GLM-5.3 (Max) and GLM-5.3-Flash appear in late-2026 LMArena with MIT-marked offerings and aggressive Flash pricing (~$0.05/$0.18 class on some Flash rows). Strengths: open-leaning licenses, strong CN enterprise footprint, agent / tool demos, cost for high volume. Limits: verify which checkpoint is MIT downloadable vs API-only. Desk tips: popular for CN ISV white-label; pin Flash vs Max in latency SLAs. Z.ai: https://z.ai/ Zhipu: https://www.zhipuai.cn/

Anonymous · 2026-10-31 22:27

Submit a complaint

Currently this platform charges no fees

Category [optional]: AI models Cluster [optional]: China / Beijing Who: Baidu. ERNIE 5.1 / 5.0 family on Qianfan and consumer Wenxin desks; mid-pack LMArena presence (~50s Elo band in Oct 2026 snapshots) with strong China distribution. Strengths: Baidu search/cloud integration, CN compliance familiarity, speech and multimodal add-ons. Limits: proprietary; international buyers should confirm language quality on non-Chinese prompts. Desk tips: use Qianfan model ids in RFQs; ask for VPC Qianfan options. ERNIE / Wenxin: https://yiyan.baidu.com/ Qianfan: https://cloud.baidu.com/product/wenxinworkshop

Anonymous · 2026-10-31 22:27

Submit a complaint

Currently this platform charges no fees

Category [optional]: AI models Cluster [optional]: China Who: DeepSeek (China). Strong open-weights + API story; late-2026 arena entries include DeepSeek-V4.1-Flash (Max), DeepSeek-V4-Pro, Flash Preview — many under MIT-class licensing for weights where published. Strengths: cost/performance, coding and math reputation, attractive for self-host and high-volume API. Limits: confirm export/compliance for your buyer jurisdiction; API endpoints and weight releases may diverge. Desk tips: great candidate for factory private cloud RFQs; always attach license file hash and eval date. Site: https://www.deepseek.com/ API: https://platform.deepseek.com/

Anonymous · 2026-10-31 22:27

Submit a complaint

Currently this platform charges no fees

Category [optional]: AI models Cluster [optional]: Canada / Global enterprise Who: Cohere. Enterprise-focused Command LLMs plus market-leading Embed and Rerank for RAG desks — less consumer chat hype, more retrieval stacks for trade knowledge bases. Strengths: RAG building blocks, VPC/on-prem narratives, multilingual embeds. Limits: not always top of public chat arenas — evaluate on your corpus. Site: https://cohere.com/

Anonymous · 2026-10-31 22:27

Submit a complaint

Currently this platform charges no fees

Category [optional]: AI models Cluster [optional]: China / Beijing–Singapore cloud paths Who: ByteDance. Consumer Doubao app; Volcengine / model APIs; arena name Dola Seed 2.0 Pro among late-2026 entries. Strengths: distribution scale (apps + ads + cloud), competitive CN pricing, multimodal roadmap. Limits: proprietary; international data-path diligence required. Volcengine: https://www.volcengine.com/ Doubao: https://www.doubao.com/

Anonymous · 2026-10-31 22:27

Submit a complaint

Currently this platform charges no fees

Category [optional]: AI models Cluster [optional]: United States Who: Anthropic (US). Claude family for chat, long-context analysis, coding, and computer/tool use. Late-2026 LMArena leaders often include Claude Opus 5.5 (High), Claude Fable 5.x, and Opus 4.x / Sonnet 5.x variants with 200K–1M context classes. Strengths: careful instruction following, long documents, coding quality, Constitutional AI / safety narrative that many enterprises cite in RFPs. Limits: proprietary API; cost of Max/High thinking modes; availability by region. Desk tips: pin model id + thinking budget in SOWs; require audit logs for agent tool calls; human approval on payments. Site: https://www.anthropic.com/ Claude: https://claude.ai/ API docs: https://docs.anthropic.com/

Anonymous · 2026-10-31 22:27

Submit a complaint

Currently this platform charges no fees

Category [optional]: AI models Cluster [optional]: United States / Global AWS Who: Amazon. Nova model family + Amazon Bedrock hosting third-party models (Anthropic, Meta, Mistral, Cohere, etc.) under AWS IAM, PrivateLink, and region controls. Strengths: enterprise procurement path, multi-model single bill, strong compliance tooling. Limits: latency/price depend on chosen model + region; Bedrock ≠ one quality rank. Desk tips: RFQs should list Bedrock model ids and inference profiles, not only “Nova”. Bedrock: https://aws.amazon.com/bedrock/ Nova: https://aws.amazon.com/ai/generative-ai/nova/

Anonymous · 2026-10-31 22:27

Submit a complaint

Currently this platform charges no fees

Category [optional]: AI models / inference Cluster [optional]: Global Foundation models, fine-tunes, LoRA packs and GPU/NPU inference. Publish license (open weights vs API), context limits and region. Named vendors only.

Anonymous · 2026-10-31 08:11

Submit a complaint