Official LLM API price list
A "0.3x rate" only means something against a baseline — this table is that baseline. Prices are taken from each vendor's own pricing page, at list price (excluding volume discounts), in USD per million tokens. Work out real cost in the rate calculator →
OpenAI
| Model | Tier | Input $/M | Output $/M | Cached input | Context | Services | Notes |
|---|---|---|---|---|---|---|---|
|
GPT-6.1 Sol
gpt-6.1-sol
|
Flagship | $2 | $10 | — | 1050K | 101 → | |
|
GPT-6 Sol
gpt-6-sol
|
Workhorse | $2 | $10 | — | 1050K | 101 → | |
|
GPT-6 Luna
cheapest in family
gpt-6-luna
|
Lightweight | $0.1 | $0.5 | — | 1050K | 101 → | |
|
GPT-5.5
gpt-5.5
|
Flagship | $5 | $30 | — | 1050K | 101 → | |
|
GPT-6 Astra
gpt-6-astra
|
Flagship | $10 | $50 | $1 | 1050K | 101 → | List price at short context ( |
|
GPT-5.6 Sol
gpt-5.6-sol
|
Flagship | $4 | $20 | $0.4 | 1050K | 101 → | List price at short context ( |
|
GPT-5.6 Terra
gpt-5.6-terra
|
Workhorse | $2 | $12 | $0.2 | 1050K | 101 → | The OpenAI workhorse tier with balanced price-performance — the default choice for most production apps. |
|
GPT-5.6 Luna
gpt-5.6-luna
|
Lightweight | $0.2 | $1.2 | $0.02 | 1050K | 101 → | Cut by 80% on 2026-07-30 |
|
GPT-5.4
gpt-5.4
|
Workhorse | $2.5 | $15 | $0.25 | 400K | 101 → | The previous-generation workhorse tier — still offered at low rate multipliers by many relay services. |
|
GPT-5.4 mini
gpt-5.4-mini
|
Lightweight | $0.75 | $4.5 | $0.075 | 400K | 101 → | A small workhorse tier for cost-sensitive, high-volume calls. |
These services offer this family: Avian, Featherless, TiMi CC, Yuegle API, api.gpt.ge — full list: GPT relay services
Anthropic
| Model | Tier | Input $/M | Output $/M | Cached input | Context | Services | Notes |
|---|---|---|---|---|---|---|---|
|
Claude Opus 5.5
claude-opus-5-5
|
Flagship | $4 | $20 | — | 1000K | 81 → | |
|
Claude Sonnet 5.5
claude-sonnet-5-5
|
Workhorse | $2 | $10 | — | 1000K | 81 → | |
|
Claude Haiku 5.5
cheapest in family
claude-haiku-5-5
|
Lightweight | $0.1 | $0.5 | — | 1000K | 81 → | |
|
Claude Fable 5
claude-fable-5
|
Reasoning | $10 | $50 | — | 1000K | 81 → | |
|
Claude Opus 5
claude-opus-5
|
Flagship | $5 | $25 | $0.5 | 1000K | 81 → | Batch 50% off; cache reads save up to 90% |
|
Claude Sonnet 5
claude-sonnet-5
|
Workhorse | $2 | $10 | $0.2 | 1000K | 81 → | Introductory price made permanent on 2026-08-10; cache read $0.20, cache write $2.50; batch 50% off |
|
Claude Haiku 4.5
claude-haiku-4-5
|
Lightweight | $1 | $5 | $0.1 | 200K | 81 → | The Claude lightweight tier: fast and cheap, suited to short outputs and high concurrency. |
|
Claude Fable 5.1
claude-fable-5-1
|
Reasoning | $10 | $50 | $0.25 | 1000K | 81 → | Cache read cut to $0.25 on 2026-09-01 |
These services offer this family: Avian, Yuegle API, api.gpt.ge, api.vveai.com, ePhone AI — full list: Claude relay services
| Model | Tier | Input $/M | Output $/M | Cached input | Context | Services | Notes |
|---|---|---|---|---|---|---|---|
|
Gemini 4 Argon
gemini-4-argon
|
Flagship | $2 | $10 | — | 1048K | 63 → | |
|
Gemini 3.1 Flash-Lite
cheapest in family
gemini-3.1-flash-lite
|
Lightweight | $0.25 | $1.5 | — | 1048K | 63 → | |
|
Gemini 3.8 Flash
gemini-3.8-flash
|
Workhorse | $0.75 | $3.75 | $0.075 | 1000K | 63 → | Promotional price through 2026-12-31; doubles from 2027-01-01 |
|
Gemini 3.1 Pro
gemini-3.1-pro
|
Flagship | $2 | $12 | $0.2 | 1000K | 63 → | Cache read $0.20; long-context requests are priced in separate tiers |
|
Gemini 3.5 Flash-Lite
gemini-3.5-flash-lite
|
Lightweight | $0.3 | $2.5 | $0.03 | 1000K | 63 → | Cache read $0.03; a free allowance is also available |
These services offer this family: Yuegle API, api.gpt.ge, api.vveai.com, ePhone AI, apipro.maynor1024.live — full list: Gemini relay services
xAI
| Model | Tier | Input $/M | Output $/M | Cached input | Context | Services | Notes |
|---|---|---|---|---|---|---|---|
|
Grok 4.3
grok-4.3
|
Workhorse | $1.25 | $2.5 | — | 1000K | 26 → | |
|
Grok 4.20
grok-4.20
|
Workhorse | $1.25 | $2.5 | — | 1000K | 26 → | |
|
Grok 4.7
grok-4.7
|
Flagship | $1.6 | $4.8 | $0.4 | 500K | 26 → | Cache read $0.40; about 20% cheaper than 4.6 |
|
Grok 4.6
grok-4.6
|
Workhorse | $2 | $6 | $0.5 | 500K | 26 → | Doubles for the portion above 200K context |
|
Grok 4.1 Fast
cheapest in family
grok-4.1-fast
|
Lightweight | $0.2 | $0.5 | — | 2000K | 26 → | The xAI fast tier — both input and output prices are very low, suited to high-throughput workloads. |
These services offer this family: Yuegle API, api.gpt.ge, api.vveai.com, 便携AI聚合API, API易 — full list: Grok relay services
Zhipu AI
| Model | Tier | Input $/M | Output $/M | Cached input | Context | Services | Notes |
|---|---|---|---|---|---|---|---|
|
GLM-5.1
glm-5.1
|
Workhorse | $0.85 | $3.38 | $0.183 | 200K | 8 → | 官方人民币价：输入 6 元 / 输出 24 元 / 缓存命中 1.3 元（[0,32K) 档） |
|
GLM-5V-Turbo
glm-5v-turbo
|
Workhorse | $0.7 | $3.1 | — | 204K | 8 → | 多模态（视觉理解）；官方 5 元 / 22 元为 [0,32K) 档 |
|
GLM-5.3
glm-5.3
|
Flagship | $1.13 | $3.94 | $0.282 | 1000K | 8 → | Official ¥8/¥28 per million tokens (cached input ¥2), converted at 1 USD = 7.1 CNY; coding and agent ability… |
|
GLM-5.3-Flash
cheapest in family
glm-5.3-flash
|
Workhorse | $0.11 | $0.39 | $0.032 | 1000K | 8 → | Official ¥0.8/¥2.8 per million tokens (cached input ¥0.23), converted at 1 USD = 7.1 CNY; multimodal, understands… |
|
GLM-5.2
glm-5.2
|
Workhorse | $1.13 | $3.94 | $0.282 | 1000K | 8 → | Official ¥8/¥28 per million tokens (cached input ¥2), converted at 1 USD = 7.1 CNY |
|
GLM-4.7
glm-4.7
|
Workhorse | $0.42 | $1.97 | $0.085 | 200K | 8 → | Official ¥3/¥14 per million tokens (cached input ¥0.6), converted at 1 USD = 7.1 CNY; the tier for input |
|
GLM-4.5-Air
glm-4.5-air
|
Lightweight | $0.11 | $0.85 | — | 128K | 8 → | Official ¥0.8/¥6 per million tokens, converted at 1 USD = 7.1 CNY; the tier for input |
These services offer this family: 稳明光语纪, 维塔斯中转站, midalo, LinoRoute, AionRouter
Moonshot AI
| Model | Tier | Input $/M | Output $/M | Cached input | Context | Services | Notes |
|---|---|---|---|---|---|---|---|
|
Kimi K2.7 Code Highspeed
kimi-k2.7-code-highspeed
|
Coding | $1.83 | $7.61 | $0.366 | 262K | 10 → | 官方 13 元 / 54 元，缓存命中 2.6 元（262K 上下文） |
|
Kimi K3
kimi-k3
|
Flagship | $2.82 | $14.08 | $0.282 | 1048K | 10 → | Official ¥20/¥100 per million tokens (cached input ¥2), converted at 1 USD = 7.1 CNY; cache write billed separately at… |
|
Kimi K2.7 Code
cheapest in family
kimi-k2.7-code
|
Coding | $0.92 | $3.8 | $0.183 | 262K | 10 → | Official ¥6.5/¥27 per million tokens (cached input ¥1.3), converted at 1 USD = 7.1 CNY |
|
Kimi K2.6
kimi-k2.6
|
Workhorse | $0.92 | $3.8 | $0.155 | 262K | 10 → | Official ¥6.5/¥27 per million tokens (cached input ¥1.1), converted at 1 USD = 7.1 CNY |
These services offer this family: BlueDoAI（小蓝助手）, 稳明光语纪, 维塔斯中转站, midalo, LinoRoute
ByteDance
| Model | Tier | Input $/M | Output $/M | Cached input | Context | Services | Notes |
|---|---|---|---|---|---|---|---|
|
Doubao-Seed-2.1-Lite
doubao-seed-2.1-lite
|
Lightweight | $0.11 | $0.38 | — | 1024K | 1 → | 官方 0.8 元 / 2.7 元（[0,1024]K 档） |
|
Doubao-Seed-Evolving
doubao-seed-evolving
|
Flagship | $0.85 | $4.23 | $0.169 | 1024K | 1 → | 官方 6 元 / 30 元，缓存命中 1.2 元（[0,1024]K 档） |
|
Doubao-Seed-2.1-Pro
doubao-seed-2.1-pro
|
Flagship | $0.85 | $4.23 | — | 256K | 1 → | Official ¥6/¥30 per million tokens, converted at 1 USD = 7.1 CNY; built for production-grade tasks |
|
Doubao-Seed-2.1-Turbo
doubao-seed-2.1-turbo
|
Workhorse | $0.42 | $2.11 | — | 256K | 1 → | Official ¥3/¥15 per million tokens, converted at 1 USD = 7.1 CNY; balances quality and cost |
|
Doubao-Seed-2.0-Pro
doubao-seed-2.0-pro
|
Workhorse | $0.45 | $2.25 | $0.09 | 256K | 1 → | Official ¥3.2/¥16 per million tokens (cached input ¥0.64), converted at 1 USD = 7.1 CNY; input tier ≤32K, doubling… |
|
Doubao-Seed-2.0-Lite
doubao-seed-2.0-lite
|
Workhorse | $0.08 | $0.51 | $0.017 | 256K | 1 → | Official ¥0.6/¥3.6 per million tokens (cached input ¥0.12), converted at 1 USD = 7.1 CNY; all-modality understanding,… |
|
Doubao-Seed-2.0-Mini
cheapest in family
doubao-seed-2.0-mini
|
Lightweight | $0.03 | $0.28 | $0.006 | 256K | 1 → | Official ¥0.2/¥2 per million tokens (cached input ¥0.04), converted at 1 USD = 7.1 CNY; low latency and low cost,… |
These services offer this family: BlueDoAI（小蓝助手）
MiniMax
| Model | Tier | Input $/M | Output $/M | Cached input | Context | Services | Notes |
|---|---|---|---|---|---|---|---|
|
MiniMax M3.1 Flash
cheapest in family
minimax-m3.1-flash
|
Lightweight | $0.14 | $0.56 | $0.014 | 1000K | — | 官方 1 元 / 4 元，缓存读取 0.1 元 |
|
MiniMax M3
minimax-m3
|
Flagship | $0.3 | $1.2 | $0.06 | 1000K | — | Official $0.30/$1.20 per million tokens (includes a permanent 50% discount); unit price doubles above 512K input;… |
|
MiniMax M2.7
minimax-m2.7
|
Workhorse | $0.3 | $1.2 | $0.06 | 204K | — | Official $0.30/$1.20 per million tokens (includes a permanent 50% discount); cached input $0.06 |
|
MiniMax M2.5
minimax-m2.5
|
Workhorse | $0.3 | $1.2 | $0.03 | 204K | — | Official $0.30/$1.20 per million tokens (includes a permanent 50% discount); cached input $0.03 |
|
MiniMax M2.1
minimax-m2.1
|
Lightweight | $0.3 | $1.2 | $0.03 | 204K | — | Official $0.30/$1.20 per million tokens (includes a permanent 50% discount); cached input $0.03 |
DeepSeek
| Model | Tier | Input $/M | Output $/M | Cached input | Context | Services | Notes |
|---|---|---|---|---|---|---|---|
|
DeepSeek V4 Pro
deepseek-v4-pro
|
Flagship | $0.66 | $1.98 | $0.022 | 1000K | 40 → | Time-of-day pricing: this is the off-peak rate; roughly double at peak |
|
DeepSeek V4.1 Flash
cheapest in family
deepseek-flash
|
Lightweight | $0.15 | $0.6 | $0.003 | 1000K | 40 → | Off-peak rate; doubles at peak hours (Mon-Fri 09:00-12:00 and 14:00-18:00 Beijing time); the former name… |
These services offer this family: Avian, Featherless, Yuegle API, api.gpt.ge, api.vveai.com — full list: DeepSeek relay services
Alibaba Cloud
| Model | Tier | Input $/M | Output $/M | Cached input | Context | Services | Notes |
|---|---|---|---|---|---|---|---|
|
Qwen3.7-Max
qwen3.7-max
|
Flagship | $1.69 | $5.07 | $0.338 | 1000K | 25 → | Official ¥12/¥36 per million tokens (cached input ¥2.4), converted at 1 USD = 7.1 CNY; the current Alibaba flagship,… |
|
Qwen3.7-Plus
qwen3.7-plus
|
Workhorse | $0.28 | $1.13 | $0.056 | 1000K | 25 → | Official ¥2/¥8 per million tokens (cached input ¥0.4), converted at 1 USD = 7.1 CNY; input tier ≤256K |
|
Qwen3.6-Plus
qwen3.6-plus
|
Workhorse | $0.28 | $1.69 | — | 1000K | 25 → | Official ¥2/¥12 per million tokens, converted at 1 USD = 7.1 CNY; input tier ≤256K |
|
Qwen3.5-Plus
qwen3.5-plus
|
Workhorse | $0.11 | $0.68 | — | 256K | 25 → | Official ¥0.8/¥4.8 per million tokens, converted at 1 USD = 7.1 CNY; input tier ≤128K |
|
Qwen3.7-Flash
cheapest in family
qwen3.7-flash
|
Lightweight | $0.03 | $0.14 | — | 1000K | 25 → | Official ¥0.24/¥0.96 per million tokens, converted at 1 USD = 7.1 CNY; input tier ≤32K; a native vision-language model |
These services offer this family: Featherless, TiMi CC, api.gpt.ge, api.vveai.com, API易 — full list: Qwen relay services
Three traps in any API price list
- A cheap input price does not mean a cheap bill. Output tokens usually cost 4–6x input, so if you are generating long text or completing code, the output column is your bill.
- Cache pricing differs by an order of magnitude. For some vendors a cache hit costs 10% of list price, so if you use a fixed system prompt or repeated documents, work out the caching benefit first.
- Promotional prices expire. Models with a promotion window noted in the remarks column may double afterwards — budget at list price for anything long-term (the Notes column flags these).
Prices change as vendors adjust them; the verification date for each row is in its notes. Check the vendor's own pricing page again before making a production decision.