$ per 1M tokens · every price links to its source
The cheapest LLM APIs, at today’s verified prices.
Verified token pricing from every major provider — sorted, filtered, and visualized. No fabricated numbers: every price links back to its official source.
LLM Price Index
$4.32 /1M blended
10 frontier models · 3:1 blend · 2026-08-08 →
Models on sale
178
198 tracked incl. 20 retired
Providers
30
across 198 tracked models
Open weights
70
of 178 current models
Every current generative model by blended $/Mtok (log scale) — labelled: cheapest, priciest, and the newest launches · ● under $1 ● $1–15 ● over $15 · hover a dot to identify it, click to open · all listed in the table below
MODELPRICEWATCH.COM · 2026-08-08
Most compared
every new flagship as it lands, plus the head-to-heads people search for — never a model its own successor has replacedMeta
Muse Spark 1.2 NEW
$1.25 in · $4.25 out
Alibaba
Qwen3.8-Max NEW
$2.00 in · $6.00 out
Thinking Machines · open
Inkling Small NEW
$0.300 in · $1.20 out
Anthropic
Claude Opus 5 NEW
$5.00 in · $25.00 out
Gemini 3.5 Flash-Lite NEW
$0.300 in · $2.50 out
Gemini 3.6 Flash NEW
$1.50 in · $7.50 out
Moonshot · open
Kimi K3 NEW
$3.00 in · $15.00 out
Alibaba
Qwen3.7-Flash NEW
$0.030 in · $0.130 out
OpenAI
GPT-5.6 Sol NEW
$5.00 in · $30.00 out
xAI
Grok 4.5 NEW
$2.00 in · $6.00 out
Anthropic
Claude Sonnet 5 NEWIntro price
$2.00 in · $10.00 out
Anthropic
Claude Fable 5
$10.00 in · $50.00 out
Today’s prices
150 models · 178 host listings · $ per 1M tokens · last scan 2026-08-08| Relative cost | Notes | |||||
|---|---|---|---|---|---|---|
| GLM-4.7-FlashFree · limited-timeZ.AI | $0 |
$0 | $0 | 128K | Open | |
| Rerank 3.5Cohere | $0.020 |
$0.020 | — | — | ||
| text-embedding-3-smallOpenAI | $0.020 |
$0.020 | — | 8K | ||
| rerank-2.5-liteVoyage AI | $0.020 |
$0.020 | — | — | ||
| voyage-4-liteVoyage AI | $0.020 |
$0.020 | — | — | ||
| GLM-OCRZ.AI | $0.030 |
$0.030 | $0.030 | 128K | Open | |
| Granite 4.0 MicroIBM | $0.041 |
$0.017 | $0.112 | 128K | Open | |
| rerank-2.5Voyage AI | $0.050 |
$0.050 | — | — | ||
| Qwen3.7-FlashNEWAlibaba | $0.055 |
$0.030 | $0.130 | 1M | ||
| Llama 3.1 8BMeta | $0.058 |
$0.050 | $0.080 | 128K | Open | |
| Llama 3.1 8B InstantGroq | $0.058 |
$0.050 | $0.080 | 128K | 840 TPSOpenDeprecated | |
| voyage-4Voyage AI | $0.060 |
$0.060 | — | — | ||
| Nova MicroAmazon | $0.061 |
$0.035 | $0.140 | 128K | ||
| Baichuan M2-32BBaichuan | $0.070 |
$0.070 | $0.070 | 33K | Open | |
| Qwen-TurboAlibaba | $0.088 |
$0.050 | $0.200 | 1M | ||
| Ministral 3 3BMistral | $0.100 |
$0.100 | $0.100 | 128K | Open | |
| Reka EdgeReka | $0.100 |
$0.100 | $0.100 | 66K | ||
| GLM-4-32B-0414Z.AI | $0.100 |
$0.100 | $0.100 | 128K | Open | |
| Nova LiteAmazon | $0.105 |
$0.060 | $0.240 | 300K | ||
| Granite Embedding 278M MultilingualIBM | $0.106 |
$0.106 | — | — | Open | |
| Granite 4 H SmallIBM | $0.107 |
$0.060 | $0.250 | 128K | Open | |
| DeepSeek V4 Flash | from $0.113 |
$0.090 | $0.180 | 1M | ||
|
DeepSeek V4 Flash is sold by 3 hosts, cheapest blended first. “vs first-party” compares each host to DeepSeek, the model maker. Full DeepSeek V4 Flash host comparison → |
||||||
| Embed 4Cohere | $0.120 |
$0.120 | — | — | ||
| voyage-4-largeVoyage AI | $0.120 |
$0.120 | — | — | ||
| voyage-multimodal-3.5Voyage AI | $0.120 |
$0.120 | — | — | ||
| GPT-OSS 20B | from $0.128 |
$0.070cached $0.035 | $0.300 | 128K | Open | |
|
GPT-OSS 20B is sold by 2 hosts, cheapest blended first. No first-party API is sold here — “vs cheapest” compares each host to the lowest blended price. Full GPT-OSS 20B host comparison → |
||||||
| text-embedding-3-largeOpenAI | $0.130 |
$0.130 | — | 8K | ||
| Qwen3 32B | from $0.130 |
$0.080 | $0.280 | 128K | Open | |
|
Qwen3 32B is sold by 3 hosts, cheapest blended first. “vs first-party” compares each host to Alibaba, the model maker. Full Qwen3 32B host comparison → |
||||||
| Qwen-FlashAlibaba | $0.138 |
$0.050 | $0.400 | 1M | ||
| Voxtral Small 24BMistral | $0.150 |
$0.100 | $0.300 | 128K | Open | |
| Llama 4 Scout | from $0.150 |
$0.100 | $0.300 | 10M | Open | |
|
Llama 4 Scout is sold by 2 hosts, cheapest blended first. “vs first-party” compares each host to Meta, the model maker. Full Llama 4 Scout host comparison → |
||||||
| GLM-4.7-FlashXZ.AI | $0.153 |
$0.070cached $0.010 | $0.400 | 128K | Open | |
| Gemini 2.5 Flash-LiteGoogle | $0.175 |
$0.100 | $0.400 | 1M | Deprecated | |
| GPT-4.1 nanoOpenAI | $0.175 |
$0.100cached $0.025 | $0.400 | 1M | Deprecated | |
| voyage-code-3Voyage AI | $0.180 |
$0.180 | — | 32K | ||
| voyage-context-3Voyage AI | $0.180 |
$0.180 | — | 32K | ||
| Jamba MiniAI21 Labs | $0.250 |
$0.200 | $0.400 | 256K | Open | |
| Hunyuan Hy3 PreviewNEWTencent | $0.257 |
$0.147cached $0.037 | $0.588 | 262K | ||
| Command R 08-2024Cohere | $0.262 |
$0.150 | $0.600 | 128K | ||
| GPT-OSS 120B | from $0.262 |
$0.150cached $0.015 | $0.600 | 128K | Open | |
|
GPT-OSS 120B is sold by 3 hosts, cheapest blended first. No first-party API is sold here — “vs cheapest” compares each host to the lowest blended price. Full GPT-OSS 120B host comparison → |
||||||
| Granite 4 H MediumIBM | $0.262 |
$0.150 | $0.600 | 128K | Open | |
| Mistral Small 4Mistral | $0.262 |
$0.150 | $0.600 | 128K | ||
| Solar Pro 3Upstage | $0.262 |
$0.150cached $0.015 | $0.600 | 131K | ||
| Llama 4 MaverickDeepInfra | $0.350 |
$0.200 | $0.800 | 1M | Open | |
| Gemma-4-31B-it-PearlTogether | $0.425 |
$0.280 | $0.860 | 128K | Open | |
| GLM-4.5-AirZ.AI | $0.425 |
$0.200cached $0.030 | $1.10 | 128K | Open | |
| CodestralMistral | $0.450 |
$0.300 | $0.900 | 256K | ||
| Codestral 2508Mistral | $0.450 |
$0.300 | $0.900 | 256K | ||
| GLM-4.6VZ.AI | $0.450 |
$0.300cached $0.050 | $0.900 | 128K | Open | |
| GPT-5.6 LunaNEW −80%OpenAI | $0.450 |
$0.200cached $0.020 | $1.20 | 1M | ||
| GPT-5.4 nanoOpenAI | $0.463 |
$0.200cached $0.020 | $1.25 | 400K | ||
| MiniMax M2.5Fireworks | $0.525 |
$0.300cached $0.030 | $1.20 | 128K | Open | |
| MiniMax M2.7 | from $0.525 |
$0.300cached $0.060 | $1.20 | 128K | Open | |
|
MiniMax M2.7 is sold by 2 hosts, cheapest blended first. “vs first-party” compares each host to MiniMax, the model maker. Full MiniMax M2.7 host comparison → |
||||||
| MiniMax M3 | from $0.525 |
$0.300cached $0.060 | $1.20 | 1M | Open | |
|
MiniMax M3 is sold by 3 hosts, cheapest blended first. “vs first-party” compares each host to MiniMax, the model maker. Full MiniMax M3 host comparison → |
||||||
| Granite 4 H LargeIBM | $0.525 |
$0.300 | $1.20 | 128K | Open | |
| Inkling SmallNEWThinking Machines | $0.525 |
$0.300cached $0.060 | $1.20 | 262K | Open | |
| DeepSeek V4 Pro | from $0.544 |
$0.435cached $0.004 | $0.870 | 1M | ||
|
DeepSeek V4 Pro is sold by 3 hosts, cheapest blended first. “vs first-party” compares each host to DeepSeek, the model maker. Full DeepSeek V4 Pro host comparison → |
||||||
| Qwen3.7 Plus | from $0.560 |
$0.320 | $1.28 | 128K | Open | |
|
Qwen3.7 Plus is sold by 3 hosts, cheapest blended first. “vs first-party” compares each host to Alibaba, the model maker. Full Qwen3.7 Plus host comparison → |
||||||
| Qwen3.6-FlashAlibaba | $0.563 |
$0.250 | $1.50 | 1M | ||
| Gemini 3.1 Flash-LiteGoogle | $0.563 |
$0.250 | $1.50 | 1M | Deprecated | |
| Qwen-PlusAlibaba | $0.600 |
$0.400 | $1.20 | 131K | ||
| Llama 3.3 70B | from $0.640 |
$0.590 | $0.790 | 128K | Open | |
|
Llama 3.3 70B is sold by 3 hosts, cheapest blended first. “vs first-party” compares each host to Meta, the model maker. Full Llama 3.3 70B host comparison → |
||||||
| GPT-4.1 miniOpenAI | $0.700 |
$0.400cached $0.100 | $1.60 | 1M | ||
| Mistral Large 3Mistral | $0.750 |
$0.500 | $1.50 | 128K | ||
| Command R 03-2024Cohere | $0.750 |
$0.500 | $1.50 | 128K | Deprecated | |
| Mistral Medium 3Mistral | $0.800 |
$0.400 | $2.00 | 128K | Deprecated | |
| Gemini 3.5 Flash-LiteNEWGoogle | $0.850 |
$0.300cached $0.030 | $2.50 | 1M | ||
| Gemini 2.5 FlashGoogle | $0.850 |
$0.300 | $2.50 | 1M | Deprecated | |
| Relace Apply 3Relace | $0.900 |
$0.800 | $1.20 | 256K | ||
| SonarPerplexity | $1.00 |
$1.00 | $1.00 | 200K | ||
| GLM-4.5Z.AI | $1.00 |
$0.600cached $0.110 | $2.20 | 128K | Open | |
| GLM-4.6Z.AI | $1.00 |
$0.600cached $0.110 | $2.20 | 200K | Open | |
| GLM-4.7Z.AI | $1.00 |
$0.600cached $0.110 | $2.20 | 128K | Open | |
| NVIDIA Nemotron 3 Ultra | from $1.05 |
$0.600cached $0.120 | $2.40 | 128K | Open | |
|
NVIDIA Nemotron 3 Ultra is sold by 2 hosts, cheapest blended first. No first-party API is sold here — “vs cheapest” compares each host to the lowest blended price. Full NVIDIA Nemotron 3 Ultra host comparison → |
||||||
| Reka FlashReka | $1.10 |
$0.800 | $2.00 | 128K | ||
| Qwen3.6-Plus | from $1.13 |
$0.500 | $3.00 | 1M | ||
|
Qwen3.6-Plus is sold by 2 hosts, cheapest blended first. “vs first-party” compares each host to Alibaba, the model maker. Full Qwen3.6-Plus host comparison → |
||||||
| Gemini 3 Flash PreviewGoogle | $1.13 |
$0.500cached $0.050 | $3.00 | 1M | Preview | |
| Kimi K2.5 | from $1.20 |
$0.600cached $0.100 | $3.00 | 256K | Open | |
|
Kimi K2.5 is sold by 2 hosts, cheapest blended first. “vs first-party” compares each host to Moonshot, the model maker. Full Kimi K2.5 host comparison → |
||||||
| Qwen3.6-27B | from $1.20 |
$0.600 | $3.00 | 128K | 500 TPSOpen | |
|
Qwen3.6-27B is sold by 2 hosts, cheapest blended first. “vs first-party” compares each host to Alibaba, the model maker. Full Qwen3.6-27B host comparison → |
||||||
| QwQ-PlusAlibaba | $1.20 |
$0.800 | $2.40 | 131K | ||
| Grok Build 0.1xAI | $1.25 |
$1.00 | $2.00 | 256K | ||
| LongCat-2.0NEWIntro priceMeituan | $1.30 |
$0.750cached $0.015 | $2.95 | 262K | Open | |
| Qwen3.5-397B-A17BAlibaba | $1.35 |
$0.600 | $3.60 | 262K | Open | |
| Nova ProAmazon | $1.40 |
$0.800 | $3.20 | 300K | ||
| Relace SearchRelace | $1.50 |
$1.00 | $3.00 | 256K | ||
| GLM-5Z.AI | $1.55 |
$1.00cached $0.200 | $3.20 | 128K | Open | |
| Grok 4.20xAI | $1.56 |
$1.25cached $0.200 | $2.50 | 256K | ||
| Grok 4.3xAI | $1.56 |
$1.25 | $2.50 | 1M | ||
| GPT-5.4 miniOpenAI | $1.69 |
$0.750cached $0.075 | $4.50 | 1M | ||
| Kimi K2.6 | from $1.71 |
$0.950cached $0.160 | $4.00 | 256K | Open | |
|
Kimi K2.6 is sold by 2 hosts, cheapest blended first. “vs first-party” compares each host to Moonshot, the model maker. Full Kimi K2.6 host comparison → |
||||||
| Kimi K2.7 Code | from $1.71 |
$0.950cached $0.190 | $4.00 | 256K | Open | |
|
Kimi K2.7 Code is sold by 3 hosts, cheapest blended first. “vs first-party” compares each host to Moonshot, the model maker. Full Kimi K2.7 Code host comparison → |
||||||
| InklingNEWThinking Machines | $1.76 |
$1.00cached $0.170 | $4.05 | 262K | Open | |
| Qwen3.7-MaxIntro price | from $1.88 |
$1.25cached $0.130 | $3.75 | 1M | ||
|
Qwen3.7-Max is sold by 2 hosts, cheapest blended first. “vs first-party” compares each host to Alibaba, the model maker. Full Qwen3.7-Max host comparison → |
||||||
| GLM-5-TurboZ.AI | $1.90 |
$1.20cached $0.240 | $4.00 | 128K | Open | |
| GLM-5V-TurboZ.AI | $1.90 |
$1.20cached $0.240 | $4.00 | 128K | Open | |
| o3-miniOpenAI | $1.93 |
$1.10cached $0.550 | $4.40 | 200K | Deprecated | |
| o4-miniOpenAI | $1.93 |
$1.10cached $0.275 | $4.40 | 200K | Deprecated | |
| Claude Haiku 4.5Anthropic | $2.00 |
$1.00cached $0.100 | $5.00 | 200K | ||
| Muse Spark 1.1NEWMeta | $2.00 |
$1.25cached $0.150 | $4.25 | 1M | ||
| Muse Spark 1.2NEWMeta | $2.00 |
$1.25cached $0.150 | $4.25 | 1M | ||
| GLM 5.1 | from $2.15 |
$1.40cached $0.260 | $4.40 | 128K | Open | |
|
GLM 5.1 is sold by 2 hosts, cheapest blended first. “vs first-party” compares each host to Z.AI, the model maker. Full GLM 5.1 host comparison → |
||||||
| GLM 5.2 −46% | from $2.15 |
$1.40cached $0.140 | $4.40 | 1M | Open | |
|
GLM 5.2 is sold by 3 hosts, cheapest blended first. “vs first-party” compares each host to Z.AI, the model maker. Full GLM 5.2 host comparison → |
||||||
| Qwen3-MaxAlibaba | $2.40 |
$1.20 | $6.00 | 262K | ||
| Qwen3.8-MaxNEWAlibaba | $3.00 |
$2.00cached $0.200 | $6.00 | 1M | ||
| Gemini 3.6 FlashNEWGoogle | $3.00 |
$1.50 | $7.50 | 1M | ||
| Mistral Medium 3.5Mistral | $3.00 |
$1.50 | $7.50 | 128K | ||
| Reka CoreReka | $3.00 |
$2.00 | $6.00 | 128K | ||
| Grok 4.5NEWxAI | $3.00 |
$2.00 | $6.00 | 500K | ||
| Gemini 3.5 FlashGoogle | $3.38 |
$1.50 | $9.00 | 1M | ||
| Kimi K2.7 Code HighSpeedMoonshot | $3.42 |
$1.90cached $0.380 | $8.00 | 262K | Open | |
| Gemini 2.5 ProGoogle | $3.44 |
$1.25 | $10.00 | 2M | Deprecated | |
| Jamba LargeAI21 Labs | $3.50 |
$2.00 | $8.00 | 256K | Open | |
| GPT-4.1OpenAI | $3.50 |
$2.00cached $0.500 | $8.00 | 1M | ||
| Sonar Deep ResearchPerplexity | $3.50 |
$2.00 | $8.00 | 200K | ||
| Sonar Reasoning ProPerplexity | $3.50 |
$2.00 | $8.00 | 200K | ||
| Claude Sonnet 5NEWIntro priceAnthropic | $4.00 |
$2.00cached $0.200 | $10.00 | 1M | ||
| Command ACohere | $4.38 |
$2.50 | $10.00 | 256K | ||
| Command R+ 08-2024Cohere | $4.38 |
$2.50 | $10.00 | 128K | ||
| Yi Large01.AI | $4.50 |
$3.00 | $9.00 | 32K | ||
| Gemini 3.1 ProGoogle | $4.50 |
$2.00cached $0.200 | $12.00 | 2M | ||
| GPT-5.6 TerraNEW −20%OpenAI | $4.50 |
$2.00cached $0.200 | $12.00 | 1M | ||
| GPT-5.3 ChatOpenAI | $4.81 |
$1.75cached $0.175 | $14.00 | 128K | ||
| GPT-5.3 CodexOpenAI | $4.81 |
$1.75cached $0.175 | $14.00 | 400K | ||
| GPT-5.2OpenAI | $4.81 |
$1.75cached $0.175 | $14.00 | 400K | ||
| GPT-5.2 CodexOpenAI | $4.81 |
$1.75cached $0.175 | $14.00 | 400K | ||
| GPT-5.2 ChatOpenAI | $4.81 |
$1.75cached $0.175 | $14.00 | 128K | ||
| Nova PremierAmazon | $5.00 |
$2.50 | $12.50 | 1M | Deprecated | |
| GPT-5.4OpenAI | $5.63 |
$2.50cached $0.250 | $15.00 | 1M | ||
| Claude Sonnet 4.5Anthropic | $6.00 |
$3.00cached $0.300 | $15.00 | 200K | ||
| Claude Sonnet 4.6Anthropic | $6.00 |
$3.00cached $0.300 | $15.00 | 1M | ||
| Kimi K3NEWMoonshot | $6.00 |
$3.00cached $0.300 | $15.00 | 1M | Open | |
| Sonar ProPerplexity | $6.00 |
$3.00 | $15.00 | 200K | ||
| Gemini 3.1 Flash Lite ImageNEWGoogle | $7.69 |
$0.250 | $30.00 | — | ||
| GPT-Realtime-2OpenAI | $9.00 |
$4.00cached $0.400 | $24.00 | 128K | ||
| Claude Opus 4.5Anthropic | $10.00 |
$5.00cached $0.500 | $25.00 | 200K | ||
| Claude Opus 4.6Anthropic | $10.00 |
$5.00cached $0.500 | $25.00 | 1M | ||
| Claude Opus 4.7Anthropic | $10.00 |
$5.00cached $0.500 | $25.00 | 1M | ||
| Claude Opus 4.8Anthropic | $10.00 |
$5.00cached $0.500 | $25.00 | 1M | ||
| Claude Opus 5NEWAnthropic | $10.00 |
$5.00cached $0.500 | $25.00 | 1M | ||
| GPT-5.5OpenAI | $11.25 |
$5.00cached $0.500 | $30.00 | 1M | ||
| GPT-5.6 SolNEWOpenAI | $11.25 |
$5.00cached $0.500 | $30.00 | 1M | ||
| Fugu UltraSakana AI | $11.25 |
$5.00cached $0.500 | $30.00 | 1M | ||
| GPT-Image-2OpenAI | $13.50 |
$8.00cached $2.00 | $30.00 | 128K | ||
| Gemini 3.1 Flash ImageGoogle | $15.38 |
$0.500 | $60.00 | — | ||
| Claude Fable 5Anthropic | $20.00 |
$10.00cached $1.00 | $50.00 | 1M | ||
| Claude Mythos 5Anthropic | $20.00 |
$10.00cached $1.00 | $50.00 | 1M | ||
| Gemini 3 Pro ImageGoogle | $31.50 |
$2.00 | $120.00 | — | ||
| GPT-5.2 ProOpenAI | $57.75 |
$21.00 | $168.00 | 400K | ||
| GPT-5.5 ProOpenAI | $67.50 |
$30.00cached $3.00 | $180.00 | 1M | ||
| GPT-5.4 ProOpenAI | $67.50 |
$30.00cached $3.00 | $180.00 | 1M | ||
Every price links to its provider's official page. Hosting providers (Groq, Together) list open-weight models at their own rates; embeddings/rerankers bill on input only (output shows —). How we verify →
MODELPRICEWATCH.COM · 2026-08-08
Best model for your use case
11 use cases · ranked by independent benchmarks & verified pricesBest LLM API for Coding
Compare LLM API pricing and capabilities for coding tasks. Independently benchmarked models for code…
Best LLM API for Chatbots
Find the best LLM API for chatbots and conversational AI. Models ranked by human conversation prefer…
Cheapest LLM APIs by Token Cost
The cheapest LLM APIs ranked by token cost. Find budget-friendly models under $1/Mtok for high-volum…
Best LLM APIs with Prompt Caching
LLM APIs that support prompt caching, ranked by cached-input price. Cache hits cost 50–99% less than…
Take the data with you
For developers
Free pricing API
All 198 tracked models — 178 current plus clearly-marked retired — as structured JSON, no key, no rate limit, CORS-enabled.
curl modelpricewatch.com/api/v1/models.jsonAPI docs →
Stay informed
Price-change newsletter
A weekly email when prices drop, rise, or a new model launches across 30 providers.
Subscribe →