All models, by maker

Every model we track, grouped by who makes it — 109 models from 14 makers, 86 of them open-weight, served across 575 priced endpoints. Search by model or maker; each model opens its spec sheet and full provider-by-provider price table.

Alibaba 26 models

qwen3.8-27b$2.44/1M52 intel12 hostsOpen qwen3.6-27b$2.3/1M37.7 intel8 hostsOpen qwen3.5-27b$1.755/1M34.6 intel6 hostsOpen qwen3.5-397b-a17b$2.73/1M34.3 intel10 hostsOpen qwen3.5-122b-a10b$2.34/1M32.9 intel5 hostsOpen qwen3.6-35b-a3b$1/1M32.1 intel10 hostsOpen qwen3.5-35b-a3b$1.14/1M29.9 intel8 hostsOpen qwen3-30b-a3b$0.241/1Mno score6 hostsOpen qwen3.5-9b$0.25/1Mno score5 hostsOpen qwen3-coder-30b-a3b$0.34/1Mno score4 hostsOpen qwen3-14b$0.36/1Mno score2 hostsOpen qwen3-32b$0.36/1Mno score2 hostsOpen qwen3-235b-a22b$0.438/1Mno score10 hostsOpen qwen3-vl-8b$0.572/1Mno score2 hostsOpen qwen3.8-flash$0.62/1Mno score3 hosts qwen3-vl-30b-a3b$0.65/1Mno score4 hostsOpen qwen-2.5-72b$0.76/1Mno score2 hostsOpen qwen3-next-80b-a3b$0.878/1Mno score5 hostsOpen qwen3-coder-next$0.92/1Mno score4 hostsOpen qwen3-vl-235b-a22b$1.08/1Mno score5 hostsOpen qwen3-vl-30b-a3b-thinking$1.29/1Mno score2 hostsOpen qwen3-coder$1.3/1Mno score5 hostsOpen qwen3-next-80b-a3b-thinking$1.35/1Mno score2 hostsOpen qwen3-235b-a22b-thinking$2.53/1Mno score3 hostsOpen qwen3-vl-235b-a22b-thinking$4.4/1Mno score2 hostsOpen qwen3.8-2.4t-a95b$8/1Mno score7 hostsOpen

Mistral AI 15 models

mistral-medium-3-5$9/1M30.4 intel1 hostOpen mistral-medium-3$2.4/1M12.5 intel1 hostProp.ministral-14b$0.4/1M11.2 intel1 hostOpen ministral-8b$0.3/1M9 intel1 hostOpen ministral-3b$0.2/1M7.1 intel1 hostOpen mistral-nemo$0.049/1Mno score4 hostsOpen mistral-small-3.2-24b$0.275/1Mno score2 hostsOpen voxtral-small-24b$0.4/1Mno score1 hostOpen mistral-small-4$0.75/1Mno score2 hostsOpen mistral-saba$0.8/1Mno score1 hostProp.codestral$1.2/1Mno score1 hostProp.mistral-large$2/1Mno score1 hostOpen mistral-medium-3.1$2.4/1Mno score1 hostProp.mixtral-8x22b$8/1Mno score1 hostOpen mistral-medium$9/1Mno score1 host

DeepSeek 11 models

deepseek-v4-flash$0.24/1M51.8 intel24 hostsOpen deepseek-v3.2$0.518/1M25.1 intel14 hostsOpen deepseek-chat-v3.1$1.2/1M21.4 intel7 hostsOpen deepseek-r1$2.65/1M20.4 intel4 hostsOpen deepseek-v3$1.21/1M14.2 intel1 hostOpen deepseek-v3.2-exp$0.68/1Mno score3 hostsOpen deepseek-v4-flash-vision-exp$0.862/1Mno score6 hostsOpen deepseek-chat-v3$1.14/1Mno score4 hostsOpen deepseek-chat$1.21/1Mno score2 hostsOpen deepseek-v3.1-terminus$1.25/1Mno score4 hostsOpen deepseek-v4$2.64/1Mno score20 hostsOpen

Z.ai 11 models

glm-5.3$4.64/1M59.5 intel23 hostsOpen glm-5.3-flash$0.309/1M57.5 intel22 hostsOpen glm-5.2$0/1M52.6 intel23 hostsOpen glm-5.1$3.77/1M41 intel16 hostsOpen glm-5$2.52/1M40.5 intel10 hostsOpen glm-4.7$2.15/1M34.5 intel6 hostsOpen glm-4.6$2.18/1M23.4 intel5 hostsOpen glm-4.7-flash$0.46/1M23.3 intel4 hostsOpen glm-4.5-air$0.98/1M16.7 intel3 hostsOpen glm-4.6v$1.2/1Mno score2 hostsOpen glm-4.5v$2.4/1Mno score2 hostsOpen

OpenAI 9 models

gpt-5.6-sol$6/1M58.9 intel3 hostsProp.gpt-5.6-terra$7/1M55 intel3 hostsProp.gpt-5.6-luna$0.7/1M51.2 intel3 hostsProp.gpt-oss-120b$0.2/1M24.1 intel16 hostsOpen gpt-oss-20b$0.12/1M15.2 intel12 hostsOpen gpt-4.1-mini$2/1M14.8 intel1 hostProp.gpt-4o$12.5/1M11.1 intel1 hostProp.gpt-4o-mini$0.75/1M6.7 intel1 hostProp.gpt-6-astra$60/1Mno score1 hostProp.

Google 7 models

gemini-3.8-flash$4.5/1M58.7 intel1 hostProp.gemini-3.6-flash$4.5/1M51.6 intel1 hostProp.gemma-4-31b$0.43/1M29.7 intel12 hostsOpen gemma-4-26b-a4b$0.4/1M26.1 intel8 hostsOpen gemma-3-27b$0.24/1M7.4 intel4 hostsOpen gemini-3.5-flash-lite$2.8/1Mno score1 hostProp.gemini-3.7-flash$4.5/1Mno score1 hostProp.

Meta 6 models

llama-4-maverick$0.896/1M14.5 intel5 hostsOpen llama-4-scout$0.4/1M10.3 intel3 hostsOpen llama-3.1-8b$0.06/1Mno score5 hostsOpen llama-3.2-3b$0.38/1Mno score2 hostsOpen llama-3.3-70b$0.42/1Mno score12 hostsOpen llama-3.1-70b$0.8/1Mno score3 hostsOpen

MiniMax 6 models

minimax-m3$1.19/1M45.4 intel11 hostsOpen minimax-m2.7$1.25/1M38.9 intel7 hostsOpen minimax-m2.5$1.22/1M34.5 intel8 hostsOpen minimax-m1$2.6/1M17.9 intel2 hostsOpen minimax-m2$1.275/1Mno score3 hostsOpen minimax-m2.1$1.5/1Mno score2 hostsOpen

Moonshot 5 models

kimi-k3$16.5/1M59.7 intel13 hostsOpen kimi-k2.6$3.119/1M45.1 intel17 hostsOpen kimi-k2.7-code$3.713/1M43 intel13 hostsOpen kimi-k2.5$2.7/1M36 intel8 hostsOpen kimi-k2-thinking$3.1/1Mno score2 hostsOpen

Anthropic 4 models

claude-opus-5$30/1M63 intel3 hostsProp.claude-fable-5$60/1M62.1 intel3 hostsProp.claude-sonnet-5$12/1M55.3 intel3 hostsProp.claude-fable-5-1$60/1Mno score3 hostsProp.

NVIDIA 4 models

nemotron-3-ultra-550b-a55b$2.7/1M38.3 intel3 hostsOpen nemotron-3-super-120b-a12b$0.485/1M25.7 intel2 hostsOpen nemotron-3.5-lightning$0.28/1M23.6 intel2 hostsOpen nemotron-3-nano-30b-a3b$0.25/1Mno score4 hostsOpen

Cohere 2 models

command-r$0.75/1Mno score1 hostOpen command-a$12.5/1Mno score1 hostOpen

xAI 2 models

grok-4.6$8/1M60.9 intel2 hostsProp.grok-4.5$8/1Mno score2 hostsProp.

IBM 1 model

granite-4.2-8b$0.25/1M19.6 intel2 hostsOpen

One card per model, cheapest blended (input + output) price across its hosts. Models on two or more hosts open a provider comparison; single-host models open their endpoint page — both carry the maker's declared spec sheet. Intelligence, where shown, is a third-party score we report but do not produce. Figures as of 2026-09-08.

See also the best value board, intelligence per dollar, the Chinese-model boards, and the methodology.