llmfind llms should run on premises

y intelligence indexx $ per 1M tokens, 800 in : 400 out, logsize output tokens/s ● api● open weights, priced by card write-off◌ new this fortnight
0204060 $0.001$0.01$0.1$1$10$100 $ / 1M tokens → ↑ intelligence index Claude Opus 5.5 · anthropic intelligence index 59.1 · $9.33 / 1MClaude Opus 5.5 GPT-6 Astra · openai intelligence index 56.9 · $23.3 / 1M GPT-6.1 Sol · openai intelligence index 54.6 · $4.67 / 1MGPT-6.1 Sol Claude Sonnet 5.5 · anthropic intelligence index 54.5 · $4.67 / 1M GPT-6.1 Terra · openai intelligence index 51.3 · $2.33 / 1MGPT-6.1 Terra Gemini 4 Argon · google intelligence index 50.4 · $4.67 / 1M Claude Fable 5.1 · anthropic intelligence index 50 · $23.3 / 1M Kimi K3.1 Pro · moonshotai · open weights intelligence index 46.5 · $4.67 / 1M GPT-6 Sol · openai intelligence index 44.2 · $4.67 / 1M Claude Opus 5 · anthropic intelligence index 44.2 · $11.7 / 1M Muse Wonder · meta intelligence index 43.6 · $9.33 / 1M Muse Spark 1.3 · meta intelligence index 36.6 · $2.25 / 1MMuse Spark 1.3 Claude Haiku 5.5 · anthropic intelligence index 36.5 · $0.233 / 1MClaude Haiku 5.5 Grok 4.7 · x-ai intelligence index 35.5 · $3.33 / 1M Hy4 · tencent · open weights intelligence index 34.6 · $0.695 / 1M Qwen3.8 Max (0902) · qwen intelligence index 33.7 · $3.33 / 1M GPT-5.6 Terra · openai intelligence index 32.3 · $5.33 / 1M GLM 5.3 · z-ai · open weights intelligence index 31.7 · $0.108 / 1M on 6× B200GLM 5.3 Grok 4.6 · x-ai intelligence index 31 · $3.33 / 1M Gemini 3.8 Flash · google intelligence index 30.8 · $1.93 / 1M MiMo-V2.6-Pro · xiaomi · open weights intelligence index 29.9 · $0.0983 / 1M on 7× B200MiMo-V2.6-Pro MiniMax M3.1 Flash · minimax · open weights intelligence index 29.1 · $0.0317 / 1M on 2× B200MiniMax M3.1 Flash DeepSeek V4.1 Flash 1017 · deepseek · open weights intelligence index 28.6 · $0.0549 / 1M on 6× B200 Step 5 Preview · stepfun · open weights intelligence index 28.5 · $0.0633 / 1M on 5× B200 Solar Pro 4 · upstage intelligence index 28.2 · $0.3 / 1M GPT-6 Luna · openai intelligence index 27.9 · $0.233 / 1M GLM 5.3 Flash · z-ai · open weights intelligence index 27.8 · $0.0411 / 1M on 3× B200 MiMo-V2.6-Flash · xiaomi · open weights intelligence index 26.7 · $0.0447 / 1M on 3× B200 Mistral Large 4 · mistralai · open weights intelligence index 26.1 · $0.16 / 1M on 7× B200 Muse Spark Open · meta · open weights intelligence index 24.9 · $0.129 / 1M on 6× B200 GPT-5.6 Luna · openai intelligence index 23.8 · $0.533 / 1M DeepSeek V4 Pro 0813 · deepseek · open weights intelligence index 23.1 · $1.63 / 1M Qwen3.8 27B · qwen · open weights intelligence index 20.1 · $0.0853 / 1M on 1× B200 MiniMax M3 · minimax · open weights intelligence index 18.9 · $0.0615 / 1M on 3× B200 Beam 1.0 501B A23B · reflection · open weights intelligence index 17.7 · $0.0706 / 1M on 4× B200 Kimi K3 · moonshotai · open weights intelligence index 17.4 · $5.46 / 1M Nemotron 3.5 Super VL · nvidia · open weights intelligence index 13.6 · $0.267 / 1M Inkling Small · thinkingmachines · open weights intelligence index 12.9 · $0.0287 / 1M on 2× B200Inkling Small DeepSeek V4 Pro 0423 · deepseek · open weights intelligence index 12.6 · $0.957 / 1M Inkling · thinkingmachines · open weights intelligence index 12.5 · $0.103 / 1M on 7× B200 Nemotron 3 Ultra · nvidia · open weights intelligence index 11.8 · $0.169 / 1M on 4× B200 Gemini 3.5 Flash Lite · google intelligence index 11.7 · $1.03 / 1M Claude Haiku 4.5 · anthropic intelligence index 7.4 · $2.33 / 1M Muse Glimmer 30B · meta · open weights intelligence index 6.3 · $0.0914 / 1M on 1× B200 Command A+ · cohere intelligence index 6.2 · $0.7 / 1M Gemini 3.1 Flash Lite Preview · google intelligence index 6.2 · $0.667 / 1M Nemotron 3.5 Lightning · nvidia · open weights intelligence index 5.5 · $0.00921 / 1M on 1× B200Nemotron 3.5 Lightning Granite 4.2 8B · ibm-granite · open weights intelligence index 5.1 · $0.027 / 1M on 1× B200 Mistral Medium 3.5 · mistralai intelligence index 4.6 · $3.5 / 1M Gemma 4 31B · google · open weights intelligence index 4.4 · $0.096 / 1M on 1× B200 Mistral Small 4 · mistralai · open weights intelligence index 4.3 · $0.0172 / 1M on 1× B200 North Mini Code · cohere · open weights intelligence index 3.4 · $0.00878 / 1M on 1× B200North Mini Code

try it — click a dot: the model opens on arena.ai (or OpenRouter chat when arena does not list it) and this prompt lands in your clipboard, paste it there:

Explain in three sentences why the sky is blue, then give me one surprising fact about it.
draw a dotclick the chart to fill in accuracy and cost. dots live in a cookie - only you see them.
new model?
modelintelligence$ / 1Mtok/sruns onsince
Claude Opus 5.5 anthropic●59.19.33 2026-09try
GPT-6 Astra openai●56.923.3 2026-09try
GPT-6.1 Sol openai●54.64.67 2026-09try
Claude Sonnet 5.5 anthropic●54.54.67 2026-09try
GPT-6.1 Terra openai●51.32.33 2026-10try
Gemini 4 Argon google●50.44.67 2026-10try
Claude Fable 5.1 anthropic●5023.3 2026-09try
Kimi K3.1 Pro moonshotai●46.54.67 api price2026-10try
GPT-6 Sol openai●44.24.67 2026-09try
Claude Opus 5 anthropic●44.211.7 2026-07try
Muse Wonder meta●43.69.33 2026-10try
Muse Spark 1.3 meta●36.62.25 2026-09try
Claude Haiku 5.5 anthropic●36.50.233 2026-10try
Grok 4.7 x-ai●35.53.33 2026-09try
Hy4 tencent●34.60.695 api price2026-11try
Qwen3.8 Max (0902) qwen●33.73.33 2026-09try
GPT-5.6 Terra openai●32.35.33 2026-07try
GLM 5.3 z-ai●31.70.108 6× B200 753B, 35B active2026-08try
Grok 4.6 x-ai●313.33 2026-08try
Gemini 3.8 Flash google●30.81.93 2026-09try
MiMo-V2.6-Pro xiaomi●29.90.0983 7× B200 1024B, 32B active2026-09try
MiniMax M3.1 Flash minimax●29.10.0317 2× B200 220B, 10B active2026-10try
DeepSeek V4.1 Flash 1017 deepseek●28.60.0549 6× B200 763B, 18B active2026-10try
Step 5 Preview stepfun●28.50.0633 5× B200 604B, 21B active2026-10try
Solar Pro 4 upstage●28.20.3 2026-08try
GPT-6 Luna openai●27.90.233 2026-09try
GLM 5.3 Flash z-ai●27.80.0411 3× B200 321B, 13B active2026-08try
MiMo-V2.6-Flash xiaomi●26.70.0447 3× B200 311B, 15B active2026-09try
Mistral Large 4 mistralai●26.10.16 7× B200 1000B, 52B active2026-10try
Muse Spark Open meta●24.90.129 6× B200 816B, 42B active2026-10try
GPT-5.6 Luna openai●23.80.533 2026-07try
DeepSeek V4 Pro 0813 deepseek●23.11.63 api price2026-08try
Qwen3.8 27B qwen●20.10.0853 1× B200 28B2026-08try
MiniMax M3 minimax●18.90.0615 3× B200 427B, 20B active2026-05try
Beam 1.0 501B A23B reflection●17.70.0706 4× B200 501B, 23B active2026-10try
Kimi K3 moonshotai●17.45.46 api price2026-07try
Nemotron 3.5 Super VL nvidia●13.60.267 api price2026-10try
Inkling Small thinkingmachines●12.90.0287 2× B200 266B, 9B active2026-07try
DeepSeek V4 Pro 0423 deepseek●12.60.957 api price2026-04try
Inkling thinkingmachines●12.50.103 7× B200 952B, 33B active2026-07try
Nemotron 3 Ultra nvidia●11.80.169 4× B200 550B, 55B active2026-06try
Gemini 3.5 Flash Lite google●11.71.03 2026-07try
Claude Haiku 4.5 anthropic●7.42.33 2025-10try
Muse Glimmer 30B meta●6.30.0914 1× B200 30B2026-08try
Command A+ cohere●6.20.7 2026-09try
Gemini 3.1 Flash Lite Preview google●6.20.667 2026-03try
Nemotron 3.5 Lightning nvidia●5.50.00921 1× B200 30B, 3B active2026-08try
Granite 4.2 8B ibm-granite●5.10.027 1× B200 9B2026-08try
Mistral Medium 3.5 mistralai●4.63.5 2026-04try
Gemma 4 31B google●4.40.096 1× B200 31B2026-04try
Mistral Small 4 mistralai●4.30.0172 1× B200 119B, 6B active2026-03try
North Mini Code cohere●3.40.00878 1× B200 30B, 3B active2026-06try
how open weights are priced

A card is written off over 3 years at 50% utilization, plus electricity at $0.2/kWh, times 1.3 for the box around it. The cheapest card (at most 8 per box) whose memory holds the fp8 weights wins. An output token costs card-hours ÷ (peak TFLOPS × 0.2 ÷ 2 × active parameters), an input token the same at 0.6. No margin, no idle box, no engineer: a floor, not a quote. Parameter counts come from Hugging Face; for mixture-of-experts models the active share is estimated.

card$GBTFLOPSW$ / h
RTX 50902000322105750.50
H10030000809897003.33
B20040000180225010004.48