LLM API cost
calculator.
Estimate monthly spend across 170 live-priced models from 26 providers — including the levers most calculators skip: reasoning tokens, cache hit rates with write premiums, and batch tiers. Rates come straight from vendor pricing pages, stamped with their verification date.
Jump to calculator Or start from a pre-configured scenario
InputsUSD · per-token list prices
Reasoning models bill hidden thinking tokens at the output rate. Cache misses are billed at the vendor's cache-write rate where one applies (e.g. Anthropic 5-min TTL, 1.25× input). Batch mode swaps both rates and turns cache modeling off.
ResultClaude Sonnet 4.6 · 1K req/mo
Monthly cost (API)
$13.50/mo
Per request$0.013
Annual projection$162
Effective blended rate$5.40 /1M tokens
Input cost$6.00/mo
Output cost $7.50/mo
Top 5 cheapest for this workload
- 1. KAT-Coder-Air V1 $0 · free
- 2. KAT-Coder-Exp 72B 1010 $0 · free
- 3. Leanstral $0 · free
- 4. GLM-4.5 Flash $0 · free
- 5. GLM-4.7 Flash $0 · free
Try common scenarios
Pre-configured use cases
Click any scenario to load it into the calculator. Costs below are computed from the current snapshot — they move when vendors reprice.
How it works
The math behind the estimate
Each model is priced with its own published levers: list input/output rates, cached-input rates, cache-write premiums where the vendor charges them, and batch-tier rates. Reasoning tokens are billed at the output rate, because that's how vendors bill them.
/* per request */
input_rate = cache_on ? hit × cached_price + (1 − hit) × write_or_base_price
: base_input_price /* batch mode: batch rates, cache off */
cost = in_tokens × input_rate / 1M
+ (out_tokens + reasoning_tokens) × output_rate / 1M
/* monthly */
monthly = cost × requests_per_month Prices are taken from each vendor's official pricing page and stamped with the date they were last verified (currently 2026-08-31). The comparison table runs your exact workload through all 170 models with each model's own discounts.
Full methodology lives at /methodology/. Found an error? Report it.
Worked examples
Three levers, real numbers
Computed from the current snapshot — recalculated on every data refresh.
Lever 01 · caching
A support bot, before and after caching
10,000 tickets/month on Claude Haiku 4.5 with a 4K-token system prompt + context. At an 80% cache hit rate the same workload costs a fraction of the uncached bill.
ModelClaude Haiku 4.5
Workload10K req · 4,000 in / 600 out
Uncached$70.00/mo
80% cache hit$43.20/mo
−38%
caching pays for the integration in week one
Lever 02 · reasoning tokens
The thinking-token multiplier
5,000 requests/month on GPT-5.5, 500 visible output tokens each. Add a typical 6K-token thinking budget and the bill multiplies — same prompt, same visible answer.
ModelGPT-5.5
Workload5K req · 2,000 in / 500 out
No reasoning$125/mo
+6K thinking$1,025/mo
×8.2
budget thinking tokens before you ship a reasoning model
Lever 03 · batch
Overnight jobs at batch rates
500K classification calls/month on Gemini 3.5 Flash. If the workload tolerates async turnaround, batch rates cut the bill roughly in half — a one-line change for most pipelines.
ModelGemini 3.5 Flash
Workload500K req · 500 in / 60 out
Real-time$645/mo
Batch API$323/mo
−50%
same model, same output — half the rate
Optimization
Five ways to cut the bill
01
Turn on prompt caching
Cached input is up to 90–99% cheaper than fresh input on most vendors. For agents with a stable system prompt, caching alone often cuts the bill by half or more — set the cache hit rate above and watch the savings row. The full math: prompt caching cost math.
02
Route by task complexity
Classification and extraction don't need a frontier model. Routing simple tasks to a budget tier (10–50× cheaper per token) and reserving flagships for reasoning-heavy work is the single biggest lever after caching. Compare tiers in the pricing directory.
03
Trim context aggressively
Every token in the window bills on every call. Summarize old turns, truncate retrieved chunks, and strip boilerplate — a 30% context cut is a 30% input-cost cut.
04
Batch the non-real-time work
Batch APIs run at roughly half price on OpenAI, Anthropic, and Google. If a workload tolerates async turnaround, flip the Batch toggle above and compare.
05
Watch the hidden line items
Reasoning tokens, cache writes, retries, and tool-call overhead don't show up in naive estimates — they show up on your invoice. We covered the common surprises in hidden LLM API costs.
All models
Your workload on every model
Per-request and monthly cost for the scenario configured above, across all 170 live-priced models. Click a column to sort; click a model for its full pricing hub.
| Model | Context | In $/1M | Out $/1M | Cached | Per request | Monthly | vs №1 |
|---|---|---|---|---|---|---|---|
| KAT-Coder-Air V1 Kwaipilot | not listed | Free | Free | — | $0 | $0 | — |
| KAT-Coder-Exp 72B 1010 Kwaipilot | not listed | Free | Free | — | $0 | $0 | — |
| Leanstral Mistral | documented elsewhere | Free | Free | — | $0 | $0 | — |
| GLM-4.5 Flash Zhipu (Z.ai / GLM) | 128K | Free | Free | Free | $0 | $0 | — |
| GLM-4.7 Flash Zhipu (Z.ai / GLM) | 200K | Free | Free | Free | $0 | $0 | — |
| Qwen3.7 Flash Alibaba (Qwen) | 1M | $0.03 | $0.13 | — | $0.00013 | $0.125 | — |
| Command R7B Cohere | 128K | $0.0375 | $0.15 | — | $0.00015 | $0.150 | — |
| Doubao Seed 1.6 Flash ByteDance (Doubao) | 256K | $0.0223 | $0.223 | $0.0045 | $0.00016 | $0.156 | — |
| Doubao Seed 2.0 Mini ByteDance (Doubao) | 256K | $0.0298 | $0.298 | $0.006 | $0.00021 | $0.208 | — |
| Ministral 3 3B Mistral | 128K | $0.10 | $0.10 | — | $0.00025 | $0.250 | — |
| Reka Edge Reka | not listed | $0.10 | $0.10 | — | $0.00025 | $0.250 | — |
| GLM-4 32B (0414, 128K) Zhipu (Z.ai / GLM) | 128K | $0.10 | $0.10 | — | $0.00025 | $0.250 | — |
| Doubao Seed 1.6 Lite ByteDance (Doubao) | 256K | $0.0446 | $0.357 | $0.009 | $0.00027 | $0.268 | — |
| Hunyuan A13B Tencent (Hunyuan) | 224K | $0.0744 | $0.298 | — | $0.0003 | $0.298 | — |
| Qwen3 VL Flash Alibaba (Qwen) | 256K | $0.05 | $0.40 | — | $0.0003 | $0.300 | — |
| GLM-4.7 FlashX Zhipu (Z.ai / GLM) | 200K | $0.07 | $0.40 | $0.01 | $0.00034 | $0.340 | — |
| Baichuan4 Air Baichuan | 32K | $0.146 | $0.146 | — | $0.00036 | $0.365 | — |
| Step 3.5 Flash StepFun | 256K | $0.104 | $0.313 | $0.0208 | $0.00036 | $0.365 | — |
| Yi Lightning 01.AI | 16K | $0.147 | $0.147 | — | $0.00037 | $0.368 | — |
| Ministral 3 8B Mistral | 128K | $0.15 | $0.15 | — | $0.00037 | $0.375 | — |
| Doubao Seed Character ByteDance (Doubao) | 128K | $0.119 | $0.298 | $0.0238 | $0.00039 | $0.387 | — |
| Qwen 3.5 Flash Alibaba (Qwen) | 1M | $0.10 | $0.40 | — | $0.0004 | $0.400 | — |
| Gemini 2.5 Flash-Lite Google | 1M | $0.10 | $0.40 | $0.01 | $0.0004 | $0.400 | — |
| MiMo-V2.5 Xiaomi MiMo | 1M | $0.14 | $0.28 | $0.0028 | $0.00042 | $0.420 | — |
| Doubao Seed 2.0 Lite ByteDance (Doubao) | 256K | $0.0893 | $0.536 | $0.0179 | $0.00045 | $0.446 | — |
| ERNIE 4.5 Turbo Baidu (ERNIE) | 128K | $0.119 | $0.476 | $0.0298 | $0.00048 | $0.476 | — |
| Ministral 3 14B Mistral | 128K | $0.20 | $0.20 | — | $0.0005 | $0.500 | — |
| Hunyuan Translation Lite Tencent (Hunyuan) | documented elsewhere | $0.149 | $0.446 | — | $0.00052 | $0.521 | — |
| Hunyuan Hy3 Tencent (Hunyuan) | 256K | $0.132 | $0.528 | $0.033 | $0.00053 | $0.528 | — |
| Qwen3.8-Flash Alibaba (Qwen) | 1M | $0.15 | $0.47 | — | $0.00053 | $0.535 | — |
| GLM-5.3-Flash Zhipu (Z.ai / GLM) | 1M | $0.15 | $0.50 | $0.03 | $0.00055 | $0.550 | — |
| KAT-Coder-Air V2.5 Kwaipilot | 256K | $0.149 | $0.595 | $0.0298 | $0.0006 | $0.595 | — |
| Command R Cohere | 128K | $0.15 | $0.60 | — | $0.0006 | $0.600 | — |
| Mistral Small 4 Mistral | 256K | $0.15 | $0.60 | $0.15 | $0.0006 | $0.600 | — |
| Solar Pro 3 Upstage | — | $0.15 | $0.60 | $0.015 | $0.0006 | $0.600 | — |
| Jamba Mini 1.7 AI21 Labs | 256K | $0.20 | $0.40 | — | $0.0006 | $0.600 | — |
| Doubao Seed Translation ByteDance (Doubao) | documented elsewhere | $0.178 | $0.536 | — | $0.00062 | $0.625 | — |
| Hunyuan Translation Tencent (Hunyuan) | documented elsewhere | $0.178 | $0.536 | — | $0.00062 | $0.625 | — |
| Qwen3 32B Alibaba (Qwen) | 131K | $0.16 | $0.64 | — | $0.00064 | $0.640 | — |
| Qwen3 8B Alibaba (Qwen) | 128K | $0.18 | $0.70 | — | $0.00071 | $0.710 | — |
| Qwen3 30B A3B Alibaba (Qwen) | 128K | $0.20 | $0.80 | — | $0.0008 | $0.800 | — |
| Qwen3 30B A3B Instruct 2507 Alibaba (Qwen) | 262K | $0.20 | $0.80 | — | $0.0008 | $0.800 | — |
| Doubao Seed 1.6 ByteDance (Doubao) | 256K | $0.119 | $1.19 | $0.0238 | $0.00083 | $0.833 | — |
| Doubao Seed 1.6 Vision ByteDance (Doubao) | 256K | $0.119 | $1.19 | $0.0238 | $0.00083 | $0.833 | — |
| Doubao Seed 1.8 ByteDance (Doubao) | 256K | $0.119 | $1.19 | $0.0238 | $0.00083 | $0.833 | — |
| Qwen3 Next 80B A3B Instruct Alibaba (Qwen) | 262K | $0.15 | $1.20 | — | $0.0009 | $0.900 | — |
| Qwen3 Next 80B A3B Thinking Alibaba (Qwen) | 262K | $0.15 | $1.20 | — | $0.0009 | $0.900 | — |
| Qwen3 235B A22B Instruct 2507 Alibaba (Qwen) | 262K | $0.23 | $0.92 | — | $0.00092 | $0.920 | — |
| GLM-4.5 Air Zhipu (Z.ai / GLM) | 128K | $0.20 | $1.10 | $0.03 | $0.00095 | $0.950 | — |
| Doubao Seed Code ByteDance (Doubao) | 128K | $0.178 | $1.19 | $0.0357 | $0.00095 | $0.952 | — |
| GPT-5.6 Luna OpenAI | 1.05M | $0.20 | $1.20 | $0.02 | $0.001 | $1.00 | — |
| Step 3.7 Flash StepFun | 256K | $0.201 | $1.21 | $0.0402 | $0.001 | $1.00 | — |
| GPT-5.4 nano OpenAI | 400K | $0.20 | $1.25 | $0.02 | $0.00103 | $1.03 | — |
| Codestral Mistral | documented elsewhere — not on pricing page | $0.30 | $0.90 | — | $0.00105 | $1.05 | — |
| MiniMax M2 MiniMax | 205K | $0.30 | $1.20 | $0.03 | $0.0012 | $1.20 | — |
| MiniMax M2.1 MiniMax | 205K | $0.30 | $1.20 | $0.03 | $0.0012 | $1.20 | — |
| MiniMax M2.5 MiniMax | 205K | $0.30 | $1.20 | $0.03 | $0.0012 | $1.20 | — |
| MiniMax M2.7 MiniMax | 205K | $0.30 | $1.20 | $0.06 | $0.0012 | $1.20 | — |
| MiniMax M3 MiniMax | 1M | $0.30 | $1.20 | $0.06 | $0.0012 | $1.20 | — |
| Solar Pro 4 Upstage | — | $0.30 | $1.20 | $0.06 | $0.0012 | $1.20 | — |
| Qwen3 VL Plus Alibaba (Qwen) | 256K | $0.20 | $1.60 | — | $0.0012 | $1.20 | — |
| KAT-Coder-Pro V1 Kwaipilot | not listed | $0.313 | $1.25 | $0.0625 | $0.00125 | $1.25 | — |
| KAT-Coder-Pro V2 Kwaipilot | not listed | $0.313 | $1.25 | $0.0625 | $0.00125 | $1.25 | — |
| Qwen3.6 Flash Alibaba (Qwen) | 1M | $0.25 | $1.50 | $0.25 | $0.00125 | $1.25 | — |
| Gemini 3.1 Flash-Lite Google | 1M | $0.25 | $1.50 | $0.025 | $0.00125 | $1.25 | — |
| MiMo-V2.5-Pro Xiaomi MiMo | 1M | $0.435 | $0.87 | $0.0036 | $0.0013 | $1.30 | — |
| Qwen3 Coder Flash Alibaba (Qwen) | 1M | $0.30 | $1.50 | — | $0.00135 | $1.35 | — |
| Qwen3 Coder Next Alibaba (Qwen) | 262K | $0.30 | $1.50 | — | $0.00135 | $1.35 | — |
| Qwen3 14B Alibaba (Qwen) | 131K | $0.35 | $1.40 | — | $0.0014 | $1.40 | — |
| DeepSeek V4 Flash DeepSeek | 1M | $0.44 | $1.32 | $0.014 | $0.00154 | $1.54 | — |
| Hunyuan T1 Vision Tencent (Hunyuan) | 28K | $0.446 | $1.34 | — | $0.00156 | $1.56 | — |
| Hunyuan TurboS Vision Tencent (Hunyuan) | 32K | $0.446 | $1.34 | — | $0.00156 | $1.56 | — |
| Hunyuan TurboS Vision Video Tencent (Hunyuan) | 24K | $0.446 | $1.34 | — | $0.00156 | $1.56 | — |
| Tencent HY Vision 1.5 Instruct Tencent (Hunyuan) | 24K | $0.446 | $1.34 | — | $0.00156 | $1.56 | — |
| Qwen3 30B A3B Thinking 2507 Alibaba (Qwen) | 262K | $0.20 | $2.40 | — | $0.0016 | $1.60 | — |
| Qwen3.7 Plus Alibaba (Qwen) | 1M | $0.40 | $1.60 | — | $0.0016 | $1.60 | — |
| Qwen3 235B A22B Thinking 2507 Alibaba (Qwen) | 262K | $0.23 | $2.30 | — | $0.00161 | $1.61 | — |
| Aya Expanse 32B Cohere | 128K | $0.50 | $1.50 | — | $0.00175 | $1.75 | — |
| Mistral Large 3 Mistral | documented elsewhere — not on pricing page | $0.50 | $1.50 | — | $0.00175 | $1.75 | — |
| Amazon Nova 2 Lite Amazon | 1M | $0.30 | $2.50 | — | $0.00185 | $1.85 | — |
| Gemini 2.5 Flash Google | 1M | $0.30 | $2.50 | $0.03 | $0.00185 | $1.85 | — |
| Gemini 3.5 Flash-Lite Google | 1M | $0.30 | $2.50 | $0.03 | $0.00185 | $1.85 | — |
| Qwen3.6 35B A3B Alibaba (Qwen) | 256K | $0.375 | $2.25 | $0.375 | $0.00188 | $1.88 | — |
| Qwen 3.5 Plus Alibaba (Qwen) | 256K | $0.40 | $2.40 | — | $0.002 | $2.00 | — |
| Doubao Seed 2.1 Turbo ByteDance (Doubao) | 256K | $0.446 | $2.23 | $0.0893 | $0.00201 | $2.01 | — |
| Baichuan-M2 Baichuan | 32K | $0.298 | $2.98 | — | $0.00208 | $2.08 | — |
| Doubao Seed 2.0 Code ByteDance (Doubao) | 256K | $0.476 | $2.38 | $0.0952 | $0.00214 | $2.14 | — |
| Doubao Seed 2.0 Pro ByteDance (Doubao) | 256K | $0.476 | $2.38 | $0.0952 | $0.00214 | $2.14 | — |
| Baichuan-M3-Plus Baichuan | 32K | $0.744 | $1.34 | — | $0.00216 | $2.16 | — |
| GLM-4.5 Zhipu (Z.ai / GLM) | 128K | $0.60 | $2.20 | $0.11 | $0.0023 | $2.30 | — |
| GLM-4.6 Zhipu (Z.ai / GLM) | 200K | $0.60 | $2.20 | $0.11 | $0.0023 | $2.30 | — |
| GLM-4.7 Zhipu (Z.ai / GLM) | 200K | $0.60 | $2.20 | $0.11 | $0.0023 | $2.30 | — |
| MiniMax M2.1 Highspeed MiniMax | 205K | $0.60 | $2.40 | $0.03 | $0.0024 | $2.40 | — |
| MiniMax M2.5 Highspeed MiniMax | 205K | $0.60 | $2.40 | $0.03 | $0.0024 | $2.40 | — |
| MiniMax M2.7 Highspeed MiniMax | 205K | $0.60 | $2.40 | $0.06 | $0.0024 | $2.40 | — |
| Qwen 3.5 122B A10B Alibaba (Qwen) | 256K | $0.40 | $3.20 | — | $0.0024 | $2.40 | — |
| Qwen3.6 Plus Alibaba (Qwen) | 1M | $0.50 | $3.00 | — | $0.0025 | $2.50 | — |
| Qwen3.8-27B Alibaba (Qwen) | 1M | $0.50 | $3.00 | — | $0.0025 | $2.50 | — |
| Sonar Perplexity | not listed | $1.00 | $1.00 | — | $0.0025 | $2.50 | — |
| ERNIE 5.1 Baidu (ERNIE) | 128K | $0.595 | $2.68 | — | $0.00253 | $2.53 | — |
| Reka Flash Reka | not listed | $0.80 | $2.00 | — | $0.0026 | $2.60 | — |
| Kimi K2.5 Moonshot (Kimi) | 262K | $0.60 | $3.00 | $0.10 | $0.0027 | $2.70 | — |
| Qwen3 235B A22B Alibaba (Qwen) | 131K | $0.70 | $2.80 | — | $0.0028 | $2.80 | — |
| QwQ Plus Alibaba (Qwen) | 131K | $0.80 | $2.40 | — | $0.0028 | $2.80 | — |
| KAT-Coder-Pro V2.5 Kwaipilot | 256K | $0.744 | $2.98 | $0.149 | $0.00298 | $2.98 | — |
| LongCat-2.0 Meituan | 1M | $0.744 | $2.98 | $0.0149 | $0.00298 | $2.98 | — |
| Qwen 3.5 397B A17B Alibaba (Qwen) | 256K | $0.60 | $3.60 | — | $0.003 | $3.00 | — |
| Qwen3.6 27B Alibaba (Qwen) | 256K | $0.60 | $3.60 | $0.60 | $0.003 | $3.00 | — |
| Grok Build 0.1 SpaceXAI | 256K | $1.00 | $2.00 | $0.20 | $0.003 | $3.00 | — |
| Gemini 3.6 Flash Google | 1M | $0.75 | $3.75 | $0.075 | $0.00337 | $3.38 | — |
| Gemini 3.7 Flash Google DeepMind | 1M | $0.75 | $3.75 | $0.075 | $0.00337 | $3.38 | — |
| GLM-5 Zhipu (Z.ai / GLM) | 200K | $1.00 | $3.20 | $0.20 | $0.0036 | $3.60 | — |
| Grok 4.20 (0309) Non-Reasoning SpaceXAI | 1M | $1.25 | $2.50 | $0.20 | $0.00375 | $3.75 | — |
| Grok 4.20 (0309) Reasoning SpaceXAI | 1M | $1.25 | $2.50 | $0.20 | $0.00375 | $3.75 | — |
| Grok 4.20 Multi-Agent (0309) SpaceXAI | 1M | $1.25 | $2.50 | $0.20 | $0.00375 | $3.75 | — |
| Grok 4.3 SpaceXAI | 1M | $1.25 | $2.50 | $0.20 | $0.00375 | $3.75 | — |
| GPT-5.4 mini OpenAI | 400K | $0.75 | $4.50 | $0.075 | $0.00375 | $3.75 | — |
| Kimi K2.6 Moonshot (Kimi) | 262K | $0.95 | $4.00 | $0.16 | $0.0039 | $3.90 | — |
| Kimi K2.7 Code Moonshot (Kimi) | 262K | $0.95 | $4.00 | $0.19 | $0.0039 | $3.90 | — |
| Sakana Namazu Sakana AI | 256K | $0.95 | $4.00 | $0.15 | $0.0039 | $3.90 | — |
| Doubao Seed 2.1 Pro ByteDance (Doubao) | 256K | $0.893 | $4.46 | $0.178 | $0.00402 | $4.02 | — |
| GLM-5 Turbo Zhipu (Z.ai / GLM) | 200K | $1.20 | $4.00 | $0.24 | $0.0044 | $4.40 | — |
| GLM-4.5 AirX Zhipu (Z.ai / GLM) | 128K | $1.10 | $4.50 | $0.22 | $0.00445 | $4.45 | — |
| Baichuan3-Turbo Baichuan | 32K | $1.79 | $1.79 | — | $0.00446 | $4.46 | — |
| Qwen3 Coder Plus Alibaba (Qwen) | 1M | $1.00 | $5.00 | — | $0.0045 | $4.50 | — |
| Claude Haiku 4.5 Anthropic | 200K | $1.00 | $5.00 | $0.10 | $0.0045 | $4.50 | — |
| DeepSeek V4 Pro DeepSeek | 1M | $1.32 | $3.96 | $0.044 | $0.00462 | $4.62 | — |
| GLM-5.1 Zhipu (Z.ai / GLM) | 200K | $1.40 | $4.40 | $0.26 | $0.005 | $5.00 | — |
| GLM-5.2 Zhipu (Z.ai / GLM) | 1M | $1.40 | $4.40 | $0.26 | $0.005 | $5.00 | — |
| GLM-5.3 Zhipu (Z.ai / GLM) | 1M | $1.40 | $4.40 | $0.26 | $0.005 | $5.00 | — |
| Baichuan-M2-Plus Baichuan | 32K | $1.49 | $4.46 | — | $0.00521 | $5.21 | — |
| Baichuan-M3 Baichuan | 32K | $1.49 | $4.46 | — | $0.00521 | $5.21 | — |
| Qwen3 Max Alibaba (Qwen) | 252K | $1.20 | $6.00 | — | $0.0054 | $5.40 | — |
| Baichuan4 Turbo Baichuan | 32K | $2.23 | $2.23 | — | $0.00558 | $5.58 | — |
| Mistral Medium 3.5 Mistral | documented elsewhere — not on pricing page | $1.50 | $7.50 | — | $0.00675 | $6.75 | — |
| Qwen3.8 Max Alibaba (Qwen) | 1M | $2.00 | $6.00 | — | $0.007 | $7.00 | — |
| Qwen3.8-2.4T-A95B Alibaba (Qwen) | 1M | $2.00 | $6.00 | — | $0.007 | $7.00 | — |
| Reka Core Reka | not listed | $2.00 | $6.00 | — | $0.007 | $7.00 | — |
| Grok 4.5 SpaceXAI | 500K | $2.00 | $6.00 | $0.30 | $0.007 | $7.00 | — |
| Grok 4.6 SpaceXAI | 500K | $2.00 | $6.00 | $0.50 | $0.007 | $7.00 | — |
| Gemini 2.5 Pro Google | 2M | $1.25 | $10.00 | $0.125 | $0.0075 | $7.50 | — |
| Gemini 3.5 Flash Google | 1M | $1.50 | $9.00 | $0.15 | $0.0075 | $7.50 | — |
| Jamba Large 1.7 AI21 Labs | 256K | $2.00 | $8.00 | — | $0.008 | $8.00 | — |
| Sonar Deep Research Perplexity | not listed | $2.00 | $8.00 | — | $0.008 | $8.00 | — |
| Sonar Reasoning Pro Perplexity | not listed | $2.00 | $8.00 | — | $0.008 | $8.00 | — |
| Qwen3.7 Max Alibaba (Qwen) | 1M | $2.50 | $7.50 | — | $0.00875 | $8.75 | — |
| GLM-4.5 X Zhipu (Z.ai / GLM) | 128K | $2.20 | $8.90 | $0.45 | $0.00885 | $8.85 | — |
| Baichuan3-Turbo (128K) Baichuan | 128K | $3.57 | $3.57 | — | $0.00893 | $8.93 | — |
| Claude Sonnet 5 Anthropic | 1M | $2.00 | $10.00 | $0.20 | $0.009 | $9.00 | — |
| Command R+ Cohere | 128K | $2.50 | $10.00 | — | $0.010 | $10.00 | — |
| GPT-5.6 Terra OpenAI | 1.05M | $2.00 | $12.00 | $0.20 | $0.010 | $10.00 | — |
| GPT-5.3-Codex OpenAI | 400K | $1.75 | $14.00 | $0.175 | $0.011 | $10.50 | — |
| GPT-5.4 OpenAI | 1.05M | $2.50 | $15.00 | $0.25 | $0.013 | $12.50 | — |
| Claude Sonnet 4.5 Anthropic | 200K | $3.00 | $15.00 | $0.30 | $0.013 | $13.50 | — |
| Claude Sonnet 4.6 Anthropic | 1M | $3.00 | $15.00 | $0.30 | $0.013 | $13.50 | — |
| Kimi K3 Moonshot (Kimi) | 1M | $3.00 | $15.00 | $0.30 | $0.013 | $13.50 | — |
| Sonar Pro Perplexity | not listed | $3.00 | $15.00 | — | $0.013 | $13.50 | — |
| GPT-5.6 Sol OpenAI | 1.05M | $4.00 | $20.00 | $0.40 | $0.018 | $18.00 | — |
| Claude Opus 4.5 Anthropic | 200K | $5.00 | $25.00 | $0.50 | $0.022 | $22.50 | — |
| Claude Opus 4.6 Anthropic | 1M | $5.00 | $25.00 | $0.50 | $0.022 | $22.50 | — |
| Claude Opus 4.7 Anthropic | 1M | $5.00 | $25.00 | $0.50 | $0.022 | $22.50 | — |
| Claude Opus 4.8 Anthropic | 1M | $5.00 | $25.00 | $0.50 | $0.022 | $22.50 | — |
| Claude Opus 5 Anthropic | 1M | $5.00 | $25.00 | $0.50 | $0.022 | $22.50 | — |
| chat-latest OpenAI | 400K | $5.00 | $30.00 | $0.50 | $0.025 | $25.00 | — |
| GPT-5.5 OpenAI | 1M | $5.00 | $30.00 | $0.50 | $0.025 | $25.00 | — |
| Fugu Ultra Sakana AI | 272K | $5.00 | $30.00 | $0.50 | $0.025 | $25.00 | — |
| Baichuan4 Baichuan | 32K | $14.88 | $14.88 | — | $0.037 | $37.20 | — |
| Claude Fable 5 Anthropic | 1M | $10.00 | $50.00 | $1.00 | $0.045 | $45.00 | — |
| GPT-5.4 Pro OpenAI | 1.05M | $30.00 | $180 | — | $0.150 | $150 | — |
| GPT-5.5 Pro OpenAI | 1.05M | $30.00 | $180 | — | $0.150 | $150 | — |
Common questions.
How the estimate works, where it can drift from your invoice, and what the levers mean.
Q · 01 How accurate are these estimates? +
The calculator multiplies your token counts by each vendor's published list prices, verified against the vendor's own pricing page (date in the header). It models caching, cache-write premiums, batch tiers, and reasoning tokens — but not negotiated enterprise discounts, regional taxes, or provider-side failures and retries. Treat results as a tight planning estimate, then measure a production sample.
Q · 02 What are reasoning tokens and why do they matter? +
Reasoning (thinking) models generate hidden chain-of-thought tokens before the visible answer, and vendors bill them at the output rate. A request that returns 500 visible tokens can quietly bill thousands of reasoning tokens on top. Most calculators ignore this entirely — it's the single most common reason real bills exceed estimates. Set the reasoning field to your model's typical thinking budget.
Q · 03 How does the cache hit-rate math work? +
Input tokens served from cache bill at the model's cached-input price. Cache misses bill at the vendor's cache-write rate where one exists — Anthropic charges 1.25× base input to write a 5-minute cache entry — otherwise at the base input rate. Formula:
input_rate = hit × cached_price + (1 − hit) × write_or_base_price. Models with no published cache pricing ignore the slider, and the hint under the field tells you.Q · 04 Can I combine batch mode with caching? +
Not in this calculator. Published cached-input prices are standard-tier prices; stacking the batch discount on top would double-count. When Batch is on we use the vendor's batch input/output rates and disable cache modeling — a conservative estimate. If a vendor documents combined batch+cache pricing, we'll model it.
Q · 05 How fresh is the pricing data? +
Every model's rates are verified against the vendor's own pricing page and stamped — the date in the page header is the snapshot's last verification date. When a vendor reprices, the whole page (presets, examples, table) recomputes on rebuild. Spotted a stale number? Report it and we re-verify within a day.
Q · 06 Why might my actual bill still differ? +
Variable context growth in agent loops, tool-call payloads, retries, and tokenizer differences between vendors (the same text can tokenize to different counts — some tokenizers run up to ~35% heavier on identical text). The calculator assumes fixed average tokens per request; measure a representative sample with your vendor's usage dashboard for the first weeks in production.
Q · 07 Can I share or export a scenario? +
Yes — Share scenario copies a URL that encodes every input, so a teammate opens the calculator in exactly your state. Export CSV downloads the full 170-model comparison for your current workload. Both are free, no signup.
Q · 08 Are the $0 models really free? +
The free-tier entries (e.g. GLM Flash tiers, Leanstral, Hunyuan Lite) publish a genuine $0 per-token API price, typically with rate limits, capacity queues, or data-use terms attached. They're real options for prototypes and low-volume tools — read the vendor's terms before betting production traffic on one.
Deep dive
Per-model pricing hubs
01.AIYi LightningAI21 LabsJamba Large 1.7Alibaba (Qwen)Qwen 3.5 122B A10BAmazonAmazon Nova 2 LiteAnthropicClaude Fable 5BaichuanBaichuan-M2Baidu (ERNIE)ERNIE 4.5 TurboByteDance (Doubao)Doubao Seed 1.6CohereAya Expanse 32BDeepSeekDeepSeek V4 FlashGoogleGemini 2.5 FlashKwaipilotKAT-Coder-Air V2.5MeituanLongCat-2.0MiniMaxMiniMax M2MistralCodestralMoonshot (Kimi)Kimi K2.5OpenAIchat-latest All providersView all models →
Embed
Put this calculator on your site
Free to embed — your readers always see current pricing. The widget links back here for the full comparison table.
More calculators
The cost toolbox
Every tool runs on the same live-pricing backbone. See the full index →