The cheapest model isn’t the cheapest model.
Cost per token is a distraction. What matters is cost per completed task, at your volume, with oversight priced in, against a person doing the same work by hand.
$4.67 / task × 10,000 / mo = $46,667 / month
$5 in / $25 out per 1M
$2 in / $10 out per 1M
$10 in / $50 out per 1M
Done by hand
8 min @ $35/hr × 10,000/mo
100%
success
Switch to Claude Sonnet 5.
$77.47 / month
saved on model cost vs Opus 4.8 (−58%) · 4.1× cheaper than by hand, all in · 10,000 tasks/mo
Method · the formula behind the verdict+
Failed calls are retried until success, so the model cost divides by p. The by hand baseline is F · minutes per task · w / 60. Success rate, output adjustment and review time are your estimates, not published benchmarks. 1 token ≈ 4 characters of English text.
Full price sheet · July 2026 snapshot+
| Provider | Model | Input $/1M | Cached input $/1M | Output $/1M |
|---|---|---|---|---|
| Claude Opus 4.8 | $5 | — | $25 | |
| Claude Sonnet 5Intro price through Aug 31, 2026 (standard $3 / $15) | $2 | — | $10 | |
| Claude Fable 5 | $10 | — | $50 | |
| Claude Haiku 4.5 | $1 | — | $5 | |
| GPT-5.5 | $5 | $0.5 | $30 | |
| GPT-5.5 Pro | $30 | — | $180 | |
| GPT-5.4 | $2.5 | $0.25 | $15 | |
| GPT-5.4 Mini | $0.75 | $0.075 | $4.5 | |
| GPT-5.4 Nano | $0.2 | $0.02 | $1.25 | |
| Gemini 3.1 ProTier up to 200K context ($4 / $18 above 200K) | $2 | — | $12 | |
| Gemini 3.5 Flash | $1.5 | — | $9 | |
| Gemini 3.1 Flash-Lite | $0.25 | — | $1.5 | |
| Grok 4.3 | $1.25 | $0.2 | $2.5 | |
| Grok 4.1 Fast | $0.2 | — | $0.5 | |
| DeepSeek V4 ProPromotional price (75% off); standard $1.74 / $3.48 | $0.435 | $0.004 | $0.87 | |
| DeepSeek V4 Flash | $0.14 | $0.003 | $0.28 | |
| Mistral Large | $2 | — | $6 | |
| Mistral Small | $0.1 | — | $0.3 |
Sources: Anthropic, OpenAI, Google Gemini, xAI, DeepSeek and Mistral published API pricing pages. List prices change without notice. Verify before you commit.