workload
Classify rows at scale
Short input, short label out, huge volume.
Volume is the whole story here. A one-cent-per-thousand delta becomes real money.
- Input per request
- 400
- Output per request
- 20
- Requests per month
- 200,000
- Tokens per month
- 84M
- Cheapest today
- $1.64
- Most expensive
- $1,000
Most of the spread comes from output tokens on this workload. Applying a 60% prompt cache to the cheapest row moves it by $0 a month.
| Model | In /1M | Out /1M | Per request | Per month | Share | Buy |
|---|---|---|---|---|---|---|
| Mistral NemoMistralcheapestopen weights | $0.019 | $0.030 | $0.000008 | $1.64 | Open | |
| GPT-OSS 20BOpenAIopen weights | $0.018 | $0.090 | $0.000009 | $1.80 | Openaff | |
| Qwen3.7 FlashAlibaba | $0.030 | $0.130 | $0.000015 | $2.92 | Open | |
| Llama 3.1 8BMetaopen weights | $0.050 | $0.080 | $0.000022 | $4.32 | Openaff | |
| DeepSeek V4 FlashDeepSeek | $0.086 | $0.171 | $0.000038 | $7.56 | Open | |
| Gemini 2.5 Flash-LiteGoogle | $0.100 | $0.400 | $0.000048 | $9.60 | Open | |
| GLM 5.3 FlashZ.ai | $0.150 | $0.500 | $0.00007 | $14.00 | Open | |
| Gemini 3.8 FlashGoogle | $0.750 | $3.75 | $0.000375 | $75.00 | Open | |
| Claude Haiku 4.5Anthropic | $1.00 | $5.00 | $0.0005 | $100 | Openaff | |
| Grok 4.3xAI | $1.25 | $2.50 | $0.00055 | $110 | Open | |
| GPT-5.1OpenAI | $1.25 | $10.00 | $0.0007 | $140 | Open | |
| Grok 4.6xAI | $2.00 | $6.00 | $0.00092 | $184 | Open | |
| GPT-5.2 CodexOpenAI | $1.75 | $14.00 | $0.00098 | $196 | Open | |
| Claude Sonnet 5Anthropic | $2.00 | $10.00 | $0.001 | $200 | Openaff | |
| Gemini 3.1 ProGoogle | $2.00 | $12.00 | $0.001 | $208 | Open | |
| GPT-5.4OpenAI | $2.50 | $15.00 | $0.0013 | $260 | Open | |
| Kimi K3Moonshot | $3.00 | $15.00 | $0.0015 | $300 | Open | |
| Claude Opus 5.5Anthropic | $4.00 | $20.00 | $0.002 | $400 | Openaff | |
| GPT-5.5OpenAI | $5.00 | $30.00 | $0.0026 | $520 | Open | |
| Claude Fable 5.1Anthropic | $10.00 | $50.00 | $0.005 | $1,000 | Openaff | |
| GPT-6 AstraOpenAI | $10.00 | $50.00 | $0.005 | $1,000 | Open |
Other workloads
Answer support ticketsSummarize documentsReview pull requestsRun an agent over a repoAnswer questions over your docsDraft docs and changelogs
Figures assume no retries, no failed calls, and no tool-use tokens. How the arithmetic works.
Prices read from 6 provider pricing pages on Sep 24, 2026. Some outbound links may be affiliate links. How that works.