Start for $0.99

The same answer costs 2,000 times more on one model than another.

Providers publish price per million tokens, which is not comparable across models. This index prices one standard answer, 1,000 input tokens plus 500 output tokens, across 126 models from 33 providers. Last updated August 20, 2026.

Headline figures

  • Cheapest answer: $0.000053 on Ling-3.0-flash (inclusionAI).
  • Median answer: $0.0019 on Kimi K2 Thinking.
  • Most expensive answer: $0.105 on Claude Opus 4.7 (Fast) (Anthropic).
  • Spread from cheapest to most expensive: about 2,000 times.

How this is calculated

One standard answer is 1,000 input tokens, roughly a prompt with a couple of paragraphs of context, and 500 output tokens, roughly a substantial reply. Every model is priced on that identical workload: (input rate x 1,000 + output rate x 500) divided by 1,000,000.

Change the mix and the ordering shifts, which is the point: a model with cheap input and dear output wins on short prompts and loses on long replies. The formula is published so you can recompute it against your own mix.

Sources and limits

  • Prices are public list rates from the OpenRouter model catalogue, in USD per million tokens, fetched August 20, 2026.
  • They exclude prompt caching discounts, batch pricing, and negotiated volume rates. Real bills at scale run lower.
  • Cost is not quality. A model 2,000 times cheaper is not 2,000 times worse, and on many everyday tasks the gap is small.
  • The credits column is what Whizi charges per message on its own plans, shown for the 115 models it meters individually. It is not a provider rate.

The full index

126 models, cheapest first. Free to reuse with attribution to the Whizi AI Model Cost Index and a link to this page. Raw data: JSON and CSV.

#ModelProviderPer answerPer 1,000 answers$ / 1M in$ / 1M outContextWhizi credits
1Ling-3.0-flashinclusionAI$0.000053$0.052$0.021$0.063262K1x
2Nex-N2-MiniNex Agi$0.000075$0.075$0.025$0.1262K1x
3Solar Pro 4Upstage$0.000090$0.090$0.03$0.12524K1x
4Qwen3.7 FlashQwen$0.000095$0.095$0.03$0.131M1x
5Granite 4.1 8BIbm Granite$0.000100$0.100$0.05$0.1131K1x
6Mercury 2.5Inception$0.000115$0.115$0.04$0.15260K1x
7Laguna XS 2.1Poolside$0.000120$0.120$0.06$0.12262K1x
8Hy-MT2-1.8BTencent$0.000133$0.133$0.044$0.1778Knot metered
9Phi 4Microsoft$0.000140$0.140$0.07$0.1416K1x
10Solar Mini 4Upstage$0.000150$0.150$0.05$0.2524K1x
11Nemotron 3.5 LightningNVIDIA$0.000180$0.180$0.08$0.2262K1x
12Laguna S 2.1Poolside$0.000180$0.180$0.09$0.181M1x
13Granite 4.2 8BIBM$0.000185$0.185$0.06$0.25131K1x
14Hy-MT2-30B-A3BTencent$0.000222$0.222$0.074$0.2958Knot metered
15Llama 3.3 70B InstructMeta$0.000260$0.260$0.1$0.32131K1x
16DeepSeek V4 Flash 0731DeepSeek$0.000280$0.280$0.14$0.281M1x
17MiMo-V2.6-FlashXiaomi$0.000280$0.280$0.14$0.281M1x
18GLM 5.3 FlashZ.ai$0.000290$0.290$0.04$0.51M1x
19Gemini 2.5 Flash LiteGoogle$0.000300$0.300$0.1$0.41M1x
20GPT-6 LunaOpenAI$0.000350$0.350$0.1$0.51M1x
21Qwen3.8 FlashQwen$0.000385$0.385$0.15$0.471M1x
22Qwen3.8 Omni FlashQwen$0.000385$0.385$0.15$0.471M1x
23Ring-2.6-1TinclusionAI$0.000388$0.388$0.075$0.625262K1x
24Hy3Tencent$0.000396$0.396$0.132$0.528262K1x
25KAT-Coder-Air V2.5Kwaipilot$0.000450$0.450$0.15$0.6256K1x
26DeepSeek V3.2DeepSeek$0.000469$0.469$0.269$0.4164K1x
27Qwen3 Coder NextQwen$0.000520$0.520$0.12$0.8262K1x
28Llama 4 MaverickMeta$0.000600$0.600$0.2$0.81M1x
29Qwen3.6 35B A3BQwen$0.000640$0.640$0.14$1262K1x
30DeepSeek V3.1DeepSeek$0.000725$0.725$0.25$0.95164K1x
31Codestral 2508Mistral$0.000750$0.750$0.3$0.9256K1x
32Qwen3.6 FlashQwen$0.000750$0.750$0.1875$1.1251M1x
33Nex-N2-ProNex Agi$0.000750$0.750$0.25$1262K1x
34MiniMax M2MiniMax$0.000765$0.765$0.255$1.02205K1x
35Step 3.7 FlashStepfun$0.000775$0.775$0.2$1.15262K1x
36GPT-5.6 Luna ProOpenAI$0.000800$0.800$0.2$1.21Mnot metered
37GPT-5.6 LunaOpenAI$0.000800$0.800$0.2$1.21M1x
38GPT-5.4 NanoOpenAI$0.000825$0.825$0.2$1.25400K2x
39MiMo-V2.6-ProXiaomi$0.000870$0.870$0.435$0.871M2x
40MiniMax M3MiniMax$0.000900$0.900$0.3$1.21M1x
41LongCat 2.0Meituan$0.000900$0.900$0.3$1.21M1x
42Perceptron Mk1Perceptron$0.000900$0.900$0.15$1.533K1x
43DeepSeek V4.1 FlashDeepSeek$0.000900$0.900$0.3$1.21M1x
44Qwen3.7 PlusQwen$0.000960$0.960$0.32$1.281M1x
45GLM 5.3 FlashXZ.ai$0.000995$0.995$0.37$1.251M2x
46Nano Banana 2 Lite (Gemini 3.1 Flash Lite Image)Google$0.0010$1.000$0.25$1.566Knot metered
47Gemini 3.1 Flash LiteGoogle$0.0010$1.000$0.25$1.51M1x
48Inkling SmallThinkingmachines$0.0010$1.050$0.45$1.2524K1x
49Command A+Cohere$0.0010$1.050$0.3$1.5192K2x
50Muse Glimmer 30BMeta$0.0011$1.100$0.35$1.5131K2x
51Qwen3.5 Plus 2026-04-20Qwen$0.0012$1.200$0.3$1.81M1x
52Mistral Large 3 2512Mistral$0.0013$1.250$0.5$1.5262K2x
53GLM 4.7Z.ai$0.0013$1.275$0.4$1.75205K2x
54Gemini 3.7 FlashGoogle$0.0013$1.313$0.375$1.8751M2x
55Mistral Medium 3.1Mistral$0.0014$1.400$0.4$2131K2x
56Aion-3.0-MiniAion Labs$0.0014$1.400$0.7$1.4131K2x
57Aion 3.5 MiniAionLabs$0.0014$1.400$0.7$1.4262K3x
58GLM 4.6Z.ai$0.0015$1.500$0.5$2205K1x
59Gemini 2.5 FlashGoogle$0.0015$1.550$0.3$2.51M2x
60Gemini 3.5 Flash LiteGoogle$0.0015$1.550$0.3$2.51M2x
61GLM 5Z.ai$0.0016$1.560$0.6$1.92205K2x
62Kimi K2 0711Moonshot$0.0017$1.720$0.57$2.3131K10x
63Seed 2.1 TurboBytedance Seed$0.0018$1.750$0.5$2.5262K2x
64Kimi K2 ThinkingMoonshot$0.0019$1.850$0.6$2.5262K10x
65R1DeepSeek$0.0019$1.950$0.7$2.564K2x
66Seed-2.0-CodeBytedance Seed$0.0020$2.000$0.5$3262K3x
67Nano Banana 2 (Gemini 3.1 Flash Image)Google$0.0020$2.000$0.5$3131Knot metered
68Grok Build 0.1xAI$0.0020$2.000$1$2256K3x
69Qwen3.8 27BQwen$0.0021$2.050$0.45$3.21M3x
70Hy4 previewTencent$0.0021$2.084$0.834$2.5011M3x
71KAT-Coder-Pro V2.5Kwaipilot$0.0022$2.220$0.74$2.96256K3x
72Nova Pro 1.0Amazon$0.0024$2.400$0.8$3.2300K3x
73Nemotron 3 UltraNVIDIA$0.0024$2.400$0.6$3.6512K3x
74Kimi K2.7 CodeMoonshot$0.0025$2.460$0.71$3.5262K10x
75GLM 5.2Z.ai$0.0025$2.484$0.966$3.0361M3x
76Grok 4.3xAI$0.0025$2.500$1.25$2.51M4x
77Gemini 3.6 FlashGoogle$0.0026$2.625$0.75$3.751M3x
78Sakana NamazuSakana$0.0029$2.950$0.95$4262K4x
79DeepSeek V4 Pro 0813DeepSeek$0.0030$2.970$1.188$3.5641M2x
80InklingThinkingmachines$0.0030$2.975$0.95$4.051M4x
81GPT-5.4 MiniOpenAI$0.0030$3.000$0.75$4.5400K4x
82Muse Spark 1.2Meta$0.0034$3.375$1.25$4.251M5x
83Muse Spark 1.1Meta$0.0034$3.375$1.25$4.251M5x
84Muse Spark 1.3Meta$0.0034$3.375$1.25$4.251M5x
85Claude Haiku 4.5Anthropic$0.0035$3.500$1$5200K4x
86GLM 5.3Z.ai$0.0036$3.600$1.4$4.41M5x
87Qwen3.7 MaxQwen$0.0037$3.688$1.475$4.4251M5x
88Grok 4.7xAI$0.0040$4.000$1.6$4.8500K6x
89Grok 4.5xAI$0.0050$5.000$2$6500K6x
90Qwen3.8 MaxQwen$0.0050$5.000$2$61M6x
91Qwen3.8 2.4T A95BQwen$0.0050$5.000$2$61M8x
92Grok 4.6xAI$0.0050$5.000$2$6500K6x
93Fugu MaxSakana$0.0050$5.000$2$61M8x
94Mistral Medium 3.5Mistral$0.0053$5.250$1.5$7.5262K6x
95Gemini 3.5 FlashGoogle$0.0060$6.000$1.5$91M8x
96Aion-3.0Aion Labs$0.0060$6.000$3$6131K10x
97Aion 3.5AionLabs$0.0060$6.000$3$6262K10x
98Claude Sonnet 5Anthropic$0.0070$7.000$2$101M10x
99GPT-6 SolOpenAI$0.0070$7.000$2$101M10x
100Claude Sonnet 5.5Anthropic$0.0070$7.000$2$101M10x
101GPT-6.1 SolOpenAI$0.0070$7.000$2$101M10x
102GLM 5.3 PrimeZ.ai$0.0072$7.200$2.8$8.81M10x
103Command ACohere$0.0075$7.500$2.5$10256K10x
104Gemini 3.1 Pro PreviewGoogle$0.0080$8.000$2$121M10x
105GPT-5.6 Terra ProOpenAI$0.0080$8.000$2$121Mnot metered
106GPT-5.6 TerraOpenAI$0.0080$8.000$2$121M4x
107Nano Banana Pro (Gemini 3 Pro Image)Google$0.0080$8.000$2$12131Knot metered
108MiMo-V2.6-Pro-UltraSpeedXiaomi$0.0087$8.700$4.35$8.71M15x
109GPT-5.4OpenAI$0.010$10.000$2.5$151M15x
110GPT-5.6 Sol ProOpenAI$0.010$10.000$2.5$151Mnot metered
111GPT-5.6 SolOpenAI$0.010$10.000$2.5$151M20x
112Qwen3.8 Max PrimeQwen$0.010$10.000$4$121M15x
113Claude Sonnet 4.6Anthropic$0.011$10.500$3$151M10x
114Claude Sonnet 4.5Anthropic$0.011$10.500$3$151M10x
115Kimi K3Moonshot$0.011$10.500$3$151M10x
116Claude Opus 5.5Anthropic$0.014$14.000$4$201M20x
117Claude Opus 5Anthropic$0.018$17.500$5$251M20x
118Claude Opus 4.8Anthropic$0.018$17.500$5$251M20x
119GPT-5.5OpenAI$0.020$20.000$5$301M20x
120GPT Chat LatestOpenAI$0.020$20.000$5$30400K25x
121Fugu UltraSakana$0.020$20.000$5$301M25x
122Fugu Ultra v2Sakana$0.020$20.000$5$301M25x
123Claude Opus 5 (Fast)Anthropic$0.035$35.000$10$501Mnot metered
124Claude Fable 5Anthropic$0.035$35.000$10$501M100x
125Claude Opus 4.8 (Fast)Anthropic$0.035$35.000$10$501Mnot metered
126Claude Opus 4.7 (Fast)Anthropic$0.105$105.000$30$1501Mnot metered

Questions

Why price an answer instead of a million tokens?

Because nobody buys a million tokens. Providers quote input and output separately at different rates, so two models can look close per million and differ by a lot on the same real task. Pricing one fixed workload makes the numbers comparable.

Which model is the best value?

Value depends on the task. The useful reading is that the middle of this table is within a few hundredths of a cent per answer, so for most everyday work the price difference between good models is negligible and the choice should be made on output quality.

How often is it updated?

Prices are refreshed from the provider catalogue and the page carries the fetch date, currently August 20, 2026. Rates in this market move often, so check the date before quoting a figure.

What are Whizi credits?

Whizi charges a message against a monthly allowance, and expensive models cost more credits than cheap ones, from 1x up to 20x. The column shows that multiplier so you can see how a provider list rate maps onto a flat subscription.

How to cite this

A dataset is only reused as often as it is easy to credit, so here is the line to copy:

Whizi AI Model Cost Index, prices as of 2026-08-20. https://whizi.io/tools/model-cost-index

Published under CC BY 4.0, which permits commercial reuse as long as the index is named and this page is linked. Writers who need a figure verified before publishing can reach the maintainers at [email protected].

Next step

Price your own text with the Token cost calculator, or compare plans on Whizi pricing.