ABCDEFGHIJKLMNOPQRSTUVWXYZAAABACADAEAF
1
ProcessData taken: Sun, 4th Jan
THIS DATA HAS NOT BEEN VALIDATED. PLEASE LEAVE A COMMENT IF YOU SPOT AN ERROR
2
TL;DR:
4,243,400,000,000 tokens generates $3,721,139 revenue with a margin of $204,663 per week for programming tools
3
4
1. Extract all the model data from https://openrouter.ai/models?categories=programming%2Ctechnology%2Cacademia%2Cscience&fmt=cards
5
2. Eyeball numbers in https://openrouter.ai/rankings (note they are different, but not a million miles off)
6
3. Get the throughput for each model from the last day of the week to understand ratio of input, reasoning and output tokens
7
4. Calculate cost per token output token (divide by 1M). Multiply output token cost by total tokens (we assume output tokens - input tokens tend to be cheaper, but lets just assume that this is the max possible number)
8
9
Model cost to OpenRouterThis assumes that Open Router has to pay for the tokens at the price listed. Taking them at their word that they "pass through the pricing of the underlying model", that they don't have bulk discount deals with the providers to enable more margin. It also based on the largest cost and where the most tokens normally are (Output tokens)
10
MarginOpen router indicate a platform fee of 5.5% "OpenRouter charges a 5.5% ($0.80 minimum) fee when you purchase credits. We pass through the pricing of the underlying model providers without any markup, so you pay the same rate as you would directly with the provider."
For the profit per model to be true, I am making the assumption that a developer sticks with one model for the week that this data is sampled from.
11
Total Reveue
Model cost (max based on output tokens) + platform fee
12
13
Note: According to dashboard, there were 5.4T tokens this week so far. The remaining 1T seem to be other models not captured in the Top 20 or in the programmign category
14
15
16
17
18
Model NameDescription
Token Throughput (model search)
Input Token CostCost per input tokenOutput Token Cost
Cost per output token
Input token throughput (1-day)Reasoning token througput (1-day)Output token throughput (1-day)
Model cost to OpenRouter based on %age of input/output
Predicted Margin
19
Mistral Small 3.2 24BOptimized for instruction following, repetition reduction, and improved function calling.22,000,000,000$0.06/M0.00000006$0.18/M0.000000182,560,000,000505,000,000$1,754.98$96.52
20
MiniMax M2.1Lightweight, state-of-the-art model optimized for coding, agentic workflows, and modern apps.73,300,000,000$0.12/M0.00000012$0.48/M0.000000489,480,000,00077,500,000174,000,000$9,477.97$521.29
21
GPT-5.2Frontier-grade model with adaptive reasoning; strong agentic and long-context performance.81,300,000,000$1.75/M0.00000175$14/M0.000014293,000,0006,970,0005,810,000$183,899.44$10,114.47
22
Claude Haiku 4.5Fast, efficient model delivering near-frontier intelligence for real-time applications.88,800,000,000$1/M0.000001$5/M0.00000512,400,000,00026,500,000328,000,000$98,672.47$5,426.99
23
MiniMax M2Compact, high-efficiency model optimized for end-to-end coding and agentic workflows.103,000,000,000$0.20/M0.0000002$1/M0.0000019,530,000,00094,000,00077,400,000$22,055.81$1,213.07
24
GPT-4o-miniAdvanced small model supporting text/image inputs; cost-effective with SOTA intelligence.110,000,000,000$0.15/M0.00000015$0.60/M0.000000615,500,000,000685,000,000$18,595.00$1,022.72
25
KAT-Coder-Pro V1 (free)KwaiKAT's most advanced agentic coding model; optimized for tool use and instruction following.111,000,000,000$0/M0$0/M06,980,000,000338,000,000$0.00$0.00
26
Devstral 2 2512 (free)Open-source model by Mistral AI specializing in agentic coding and codebase exploration.119,000,000,000$0/M0$0/M019,400,000,000456,000,000$0.00$0.00
27
Gemini 3 Pro PreviewFlagship frontier model for high-precision multimodal reasoning and agentic workflows.120,000,000,000$2/M0.000002$12/M0.00001216,700,000,0001,260,000,000726,000,000$367,539.33$20,214.66
28
gpt-oss-120bOpen-weight MoE model designed for high-reasoning, agentic, and production use cases.123,000,000,000$0.02/M0.00000002$0.10/M0.00000018,720,000,0001,180,000,000498,000,000$4,047.95$222.64
29
Gemini 2.5 Flash LiteLightweight reasoning model optimized for ultra-low latency and cost efficiency.182,000,000,000$0.10/M0.0000001$0.40/M0.000000423,300,000,000150,000,0003,250,000,000$25,152.81$1,383.40
30
Claude Opus 4.5Frontier reasoning model optimized for complex software engineering and agentic workflows.197,000,000,000$5/M0.000005$25/M0.00002543,200,000,000122,000,000707,000,000$1,059,184.29$58,255.14
31
Grok 4.1 FastAgentic tool-calling model designed for customer support and deep research.216,000,000,000$0.20/M0.0000002$0.50/M0.000000525,700,000,0002,350,000,0001,050,000,000$50,771.13$2,792.41
32
Gemini 3 Flash PreviewHigh-speed thinking model for agentic workflows, multi-turn chat, and coding assistance.312,000,000,000$0.50/M0.0000005$3/M0.00000342,400,000,000419,000,0001,840,000,000$195,454.98$10,750.02
33
Qwen3 VL 235B A22B InstructOpen-weight multimodal model unifying text generation with visual understanding.352,000,000,000$0.12/M0.00000012$0.56/M0.00000056398,000,00042,000,000$57,024.00$3,136.32
34
Claude Sonnet 4.5Advanced model optimized for real-world agents, coding workflows, and autonomous operation.389,000,000,000$3/M0.000003$15/M0.00001546,300,000,00058,400,0001,110,000,000$1,281,899.41$70,504.47
35
DeepSeek V3.2Harmonizes high computational efficiency with strong reasoning and agentic tool-use.395,000,000,000$0.112/M0.000000112$0.16/M0.0000001647,500,000,000527,000,0003,090,000,000$45,581.60$2,506.99
36
MiMo-V2-Flash (free)Open-source MoE model by Xiaomi; excels at reasoning, coding, and agent scenarios.396,000,000,000$0/M0$0/M043,000,000,0004,530,000,0002,130,000,000$0.00$0.00
37
Gemini 2.5 FlashWorkhorse model designed for advanced reasoning, coding, math, and scientific tasks.398,000,000,000$0.30/M0.0000003$2.50/M0.000002541,700,000,000660,000,0003,170,000,000$193,055.79$10,618.07
38
Grok Code Fast 1Speedy and economical reasoning model that excels at agentic coding.455,000,000,000$0.20/M0.0000002$1.50/M0.000001559,600,000,0001,440,000,000214,000,000$106,971.87$5,883.45
39
4,243,400,000,000$3,721,138.82$204,662.63
40
41
42
43
Rankings reported number5,510,000,000,000
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100