Fathom Layer / Prompt cost planning
Small prompts.
Monthly consequences.
Set the input, response length and request volume. Inspect how recorded API rates translate into a modeled bill before comparing the trade-offs.
Input limits keep scenarios finite. They do not establish a model's supported context window or account limits.
A workload.
A range of recorded rates.
GPT-5.6 Luna · OpenAI
Lowest arithmetic cost among these recorded rates. This is not a recommendation or a comparison of model capabilities.
USD, using each record's stated rate basis. Cached input, batch discounts, tool calls, taxes and other billing conditions are outside this simplified estimate.
14 recorded rates compared. Lowest modeled monthly cost $124.10.
The rates behind the result.
Recorded prices, not live quotes.
Open each record for its source and conditions.
| Model / price record | Input / 1M | Output / 1M | Per run | Per month |
|---|---|---|---|---|
| GPT-5.6 LunaOpenAI | $0.2 | $1.2 | $0.0012 | $124.10 |
DeepSeek V4.1 FlashDeepSeekPrice recordUSD · Peak hours, cache miss (off-peak is half) Record checked . Official pricing (opens in a new tab) | $0.3 | $1.2 | $0.0012 | $124.71 |
| Gemini 3.5 Flash-LiteGoogle | $0.3 | $2.5 | $0.0026 | $257.83 |
Gemini 3.8 FlashGooglePrice recordUSD · Paid tier, through Dec 31, 2026 Record checked . Official pricing (opens in a new tab) | $0.75 | $3.75 | $0.0039 | $388.58 |
DeepSeek V4 ProDeepSeekPrice recordUSD · Peak hours, cache miss (off-peak is half) Record checked . Official pricing (opens in a new tab) | $1.32 | $3.96 | $0.0041 | $413.56 |
| Claude Haiku 4.5Anthropic | $1 | $5 | $0.0052 | $518.10 |
Gemini 2.5 ProGooglePrice recordUSD · Paid tier, prompts up to 200k tokens Record checked . Official pricing (opens in a new tab) | $1.25 | $10 | $0.0103 | $1,031.63 |
| Claude Sonnet 5Anthropic | $2 | $10 | $0.0104 | $1,036.20 |
| GPT-5.6 TerraOpenAI | $2 | $12 | $0.0124 | $1,241.00 |
Gemini 3.1 Pro PreviewGooglePrice recordUSD · Paid tier, prompts up to 200k tokens Record checked . Official pricing (opens in a new tab) | $2 | $12 | $0.0124 | $1,241.00 |
| GPT-5.6 SolOpenAI | $4 | $20 | $0.0207 | $2,072.40 |
| Claude Opus 5Anthropic | $5 | $25 | $0.0259 | $2,590.50 |
| GPT-6 AstraOpenAI | $10 | $50 | $0.0518 | $5,181.00 |
| Claude Fable 5.1Anthropic | $10 | $50 | $0.0518 | $5,181.00 |
These records were checked on the dates shown, and may have changed. Confirm the linked provider rate and billing conditions before budgeting. The same token counts do not establish equivalent output quality, tokenization or capabilities.
Read the assumptions
A cost comparison
needs context.
The estimate multiplies input tokens by the recorded input rate, output tokens by the recorded output rate, then the combined per-run cost by monthly runs. USD throughout.
Rates retain the catalog's observation date and billing conditions. They have not been rechecked during this design update. Open the provider source before treating an estimate as a purchasing decision.
Models differ in quality, latency, capabilities and context limits. Sorting by modeled cost does not establish equivalent performance or a recommendation.
Cached input, batching, tool calls, taxes and other billing rules are excluded. A four-character text shortcut is a rough English token estimate; use a provider-specific count when accuracy matters.
Explore cloud versus local costs ↗