workload
Review pull requests
Diff plus repo context in, findings out.
A real diff with surrounding files regularly passes 20k input tokens.
- Input per request
- 25k
- Output per request
- 900
- Requests per month
- 400
- Tokens per month
- 10.36M
- Cheapest today
- $0.2008
- Most expensive
- $118
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 | $0.000502 | $0.2008 | Open | |
| GPT-OSS 20BOpenAIopen weights | $0.018 | $0.090 | $0.000531 | $0.2124 | Openaff | |
| Qwen3.7 FlashAlibaba | $0.030 | $0.130 | $0.000867 | $0.3468 | Open | |
| Llama 3.1 8BMetaopen weights | $0.050 | $0.080 | $0.0013 | $0.5288 | Openaff | |
| DeepSeek V4 FlashDeepSeek | $0.086 | $0.171 | $0.0023 | $0.9216 | Open | |
| Gemini 2.5 Flash-LiteGoogle | $0.100 | $0.400 | $0.0029 | $1.14 | Open | |
| GLM 5.3 FlashZ.ai | $0.150 | $0.500 | $0.0042 | $1.68 | Open | |
| Gemini 3.8 FlashGoogle | $0.750 | $3.75 | $0.0221 | $8.85 | Open | |
| Claude Haiku 4.5Anthropic | $1.00 | $5.00 | $0.0295 | $11.80 | Openaff | |
| Grok 4.3xAI | $1.25 | $2.50 | $0.0335 | $13.40 | Open | |
| GPT-5.1OpenAI | $1.25 | $10.00 | $0.0403 | $16.10 | Open | |
| Grok 4.6xAI | $2.00 | $6.00 | $0.0554 | $22.16 | Open | |
| GPT-5.2 CodexOpenAI | $1.75 | $14.00 | $0.0564 | $22.54 | Open | |
| Claude Sonnet 5Anthropic | $2.00 | $10.00 | $0.059 | $23.60 | Openaff | |
| Gemini 3.1 ProGoogle | $2.00 | $12.00 | $0.0608 | $24.32 | Open | |
| GPT-5.4OpenAI | $2.50 | $15.00 | $0.076 | $30.40 | Open | |
| Kimi K3Moonshot | $3.00 | $15.00 | $0.0885 | $35.40 | Open | |
| Claude Opus 5.5Anthropic | $4.00 | $20.00 | $0.118 | $47.20 | Openaff | |
| GPT-5.5OpenAI | $5.00 | $30.00 | $0.152 | $60.80 | Open | |
| Claude Fable 5.1Anthropic | $10.00 | $50.00 | $0.295 | $118 | Openaff | |
| GPT-6 AstraOpenAI | $10.00 | $50.00 | $0.295 | $118 | Open |
Other workloads
Answer support ticketsSummarize documentsRun an agent over a repoClassify rows at scaleAnswer 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.