AI Model Pricing Directory
Browse API pricing for 90 active AI models from OpenAI, Anthropic, Google, DeepSeek, Qwen, Kimi & more. Compare input/output costs, context windows, and capabilities.
Last updated: 2026-11-20 · 90 models from 16 providers
90
Models
16
Providers
$0.00
Cheapest Input/M
1.1M
Max Context
Need to estimate costs?
Use our interactive Pricing Calculator to compare models side-by-side and estimate monthly costs.
| Model | Input / 1M | Output / 1M | Context |
|---|---|---|---|
| GPT-6.2 Astra OpenAI's flagship model for the hardest projects, released 2026-11-17. Max input is 922K tokens; prompts above 272K input tokens bill the entire request at long-context rates. The listed cache-write price is for 30-minute writes (12-hour writes cost $20/M). Batch is 50% off, Fast mode costs 2x, and Ultrafast costs 3x Standard through 2027-02-28, then 6x. | $8.00 | $40.00 | 1.1M |
| GPT-6.2 Sol OpenAI's go-to model for everyday work, from coding to agents at scale, released 2026-11-17 at the same price as GPT-6.1 Sol. Max input is 922K tokens; prompts above 272K input tokens bill the entire request at long-context rates. The listed cache-write price is for 30-minute writes (12-hour writes cost $5/M). Batch is 50% off, Fast mode costs 2x, and Ultrafast costs 3x Standard through 2027-02-28, then 6x. | $2.00 | $10.00 | 1.1M |
| GPT-6.1 Sol Current balanced GPT-6 model for coding and professional work. Cache reads cost half as much as GPT-6 Sol. Prompts above 272K input tokens bill the entire request at long-context rates; Batch/Flex cost 50% of Standard and Fast mode costs 2x. Tool calling requires the Responses API. | $2.00 | $10.00 | 1.1M |
| GPT-6.1 Terra Fast, efficient GPT-6.1 model for visual reasoning, computer use and natural writing, at half the price of GPT-6.1 Sol. Max input is 922K tokens. Prompts above 272K input tokens bill the entire request at long-context rates (2x input, 1.5x output). The listed cache-write price is for 30-minute retention; 12-hour retention costs $2.50/M. | $1.00 | $5.00 | 1.1M |
| GPT-6 Astra OpenAI's highest-capability GPT-6 model for hard end-to-end work. Prompts above 272K input tokens bill the entire request at long-context rates; Batch and Flex cost 50% of Standard, while Fast mode costs 2x. Ultrafast mode costs 6x Standard: $60/$300 at short context or $120/$450 above 272K input tokens. | $10.00 | $50.00 | 1.1M |
| GPT-6 Sol GPT-6 model for complex coding and agentic workflows. Prompts above 272K input tokens bill the entire request at long-context rates; Batch and Flex cost 50% of Standard, while Fast mode costs 2x. | $2.00 | $10.00 | 1.1M |
| GPT-6 Luna GPT-6 model for focused, repeatable, high-volume work. Prompts above 272K input tokens bill the entire request at long-context rates; Batch and Flex cost 50% of Standard, while Fast mode costs 2x. | $0.1 | $0.5 | 1.1M |
| GPT-5.6 Sol Previous-generation flagship. The gpt-5.6 alias routes to Sol. Promotional pricing is available until 2026-11-30; explicit cache writes cost 1.25x uncached input and requests above 272K input tokens use long-context pricing. | $4.00 | $20.00 | 1.1M |
| GPT-5.6 Terra Balanced GPT-5.6 production model at OpenAI's reduced August 2026 price. Explicit cache writes cost 1.25x uncached input; requests above 272K input tokens use long-context pricing. | $2.00 | $12.00 | 1.1M |
| GPT-5.6 Luna High-volume GPT-5.6 routing model at OpenAI's reduced August 2026 price, with the same 1.05M context window and 128K maximum output as Sol and Terra. | $0.2 | $1.20 | 1.1M |
| GPT-5.4 Pro Higher-compute GPT-5.4 variant for smarter, more precise answers. Responses API only. Prompts above 272K input tokens bill the whole request at $60/$270 per 1M. No cached-input discount is listed. | $30.00 | $180.00 | 1.1M |
| GPT-5.5 Previous OpenAI frontier model for complex reasoning, coding, and professional work. GPT-6.2 Astra is the current flagship; prompts above 272K input tokens use higher long-context pricing. | $5.00 | $30.00 | 1.1M |
| GPT-5.5 Pro Higher-compute GPT-5.5 variant for the hardest professional tasks. Prompts above 272K input tokens bill at $60/$270 per 1M. No cached-input discount is listed. | $30.00 | $180.00 | 1.1M |
| GPT-5.4 Legacy OpenAI model for coding and professional work. GPT-6.1 Terra ($1/$5) and GPT-6.1 Sol ($2/$10) now cost less per token. | $2.50 | $15.00 | 1.1M |
| GPT-5.4 Mini Current lower-cost GPT-5.4 production model. | $0.75 | $4.50 | 400K |
| Model | Input / 1M | Output / 1M | Context |
|---|---|---|---|
| Claude Fable 5.2 Successor to Fable 5.1 at 20% lower rates. The listed cache-write price is for a 5-minute TTL; 1-hour writes cost $16/M, Batch input/output costs $4/$20, and Fast mode costs $16/$80. The full 1M context is billed at standard rates. | $8.00 | $40.00 | 1.0M |
| Claude Fable 5.1 Anthropic's highest-capability broadly available model for long-horizon agentic work. Cache reads cost 0.025x base input; the listed write price is for a 5-minute TTL and 1-hour writes cost $20/M. | $10.00 | $50.00 | 1.0M |
| Claude Opus 5.5 Current default Claude model for long-running agentic coding and knowledge work. Adaptive thinking is always on. The listed cache-write price is for a 5-minute TTL; 1-hour writes cost $8/M, Batch input/output costs $2/$10, and Fast mode costs $8/$40. | $4.00 | $20.00 | 1.0M |
| Claude Fable 5 Claude model for fiction and long-form creative work. Global API access was restored on 2026-07-01. The cache-write price shown is for a 5-minute TTL; 1-hour writes cost $20/M. | $10.00 | $50.00 | 1.0M |
| Claude Opus 5 Previous-generation Opus model for complex coding, agent, and knowledge-work tasks. The cache-write price shown is for a 5-minute TTL; 1-hour writes cost $10/M. | $5.00 | $25.00 | 1.0M |
| Claude Sonnet 5.5 Current balanced Claude model with the same base input/output rates as Sonnet 5, but half the cache-read price ($0.10/M). Five-minute cache writes cost $2.50/M; one-hour writes cost $4/M. The full 1M context window uses standard rates; Batch input/output are $1/$5. | $2.00 | $10.00 | 1.0M |
| Claude Haiku 5.5 Fastest current Claude model for classification, extraction and routing. Prompts up to 100K tokens cost $0.10/$0.50 per 1M; prompts above 100K bill at $0.50/$2.50. Batch is 50% off. Anthropic does not publish a release date on the pricing page; first listed on DevTk.AI on 2026-10-25. | $0.1 | $0.5 | 1.0M |
| Claude Sonnet 5 Anthropic made the launch price permanent instead of applying the previously scheduled 2026-09-01 increase. The listed cache-write price is for a 5-minute TTL; 1-hour writes cost $4/M. | $2.00 | $10.00 | 1.0M |
| Claude Opus 4.8 Previous-generation Opus model. Full 1M context remains available at standard pricing. | $5.00 | $25.00 | 1.0M |
| Claude Opus 4.6 Previous-generation Opus model. Full 1M context is available at standard pricing. | $5.00 | $25.00 | 1.0M |
| Claude Sonnet 4.6 Previous-generation balanced Claude model. Full 1M context remains available at standard pricing. | $3.00 | $15.00 | 1.0M |
| Claude Haiku 4.5 Fastest Claude model. 200K context, 64K max output. | $1.00 | $5.00 | 200K |
| Model | Input / 1M | Output / 1M | Context |
|---|---|---|---|
| Gemini 3.8 Flash Current GA Gemini Flash flagship for long-horizon coding, autonomous agents, and enterprise workflows. The 2026 introductory rate also applies to Batch and Flex at half the listed synchronous rate. | $0.75 | $3.75 | 1.0M |
| Gemini 4 Carbon Google's most capable Gemini reasoning model, aimed at complex software engineering, math and science, and autonomous multi-step agents. Default output limit is 65,536 tokens (long decode allows more). Same introductory pricing as Gemini 4 Argon, doubling on 2026-12-01. Not available on the Gemini API free tier. | $2.00 | $10.00 | 1.0M |
| Gemini 4 Neon Gemini 4 model for complex software engineering, knowledge work and science workflows, released November 2026. Unified 1M-token context; output defaults to 65,536 tokens and reaches 1M with long decode. Implicit caching only ($0.15/M cached input). Batch and Flex cost $1.50/$7.50, Priority $5.40/$27. Not available on the free tier; thinking tokens are billed as output. | $3.00 | $15.00 | 1.0M |
| Gemini 4 Hydrogen Stable Gemini 4 model for everyday coding and agentic workflows, priced below Gemini 4 Argon and Carbon. Implicit caching only ($0.075/M cached input); explicit caching is not supported. Batch and Flex cost $0.75/$3.75, Priority $2.70/$13.50. Available on the Gemini API free tier. Default output limit is 65,536 tokens (long decode allows up to 1M). | $1.50 | $7.50 | 1.0M |
| Gemini 4 Argon Google flagship reasoning model, generally available in the Gemini API since 2026-10-13. Introductory pricing doubles on 2026-12-01. Not available on the Gemini API free tier. | $2.00 | $10.00 | 1.0M |
| Gemini 3.6 Flash Previous GA Gemini Flash model. Google applies the same introductory 2026 price as Gemini 3.7 Flash; the standard rate returns on 2027-01-01. | $0.75 | $3.75 | 1.0M |
| Gemini 3.5 Flash-Lite Current GA low-cost Gemini model for high-throughput multimodal workloads. | $0.3 | $2.50 | 1.0M |
| Gemini 3.1 Pro Preview Preview Gemini Pro model. Inputs above 200K tokens use higher long-context pricing. | $2.00 | $12.00 | 1.0M |
| Gemini 3.1 Flash-Lite Low-cost multimodal Gemini model. Audio input costs $0.50/M ($0.05/M cached). Google shuts it down on 2027-05-07 and names Gemini 3.5 Flash-Lite as the replacement. | $0.25 | $1.50 | 1.0M |
| Gemini 2.5 Flash-Lite | $0.1 | $0.4 | 1.0M |
| Model | Input / 1M | Output / 1M | Context |
|---|---|---|---|
| Grok 4.7 Current xAI frontier model for coding, agents, and knowledge work. xAI publishes no text-output limit. Prompts at or above 200K tokens bill all tokens at long-context rates; the public API does not offer the 2x-priced Fast variant. | $2.00 | $6.00 | 500K |
| Grok 4.6 Previous xAI frontier reasoning model with low, medium, high, and xhigh effort. xAI publishes no text-output limit. Prompts at or above 200K tokens bill all tokens at long-context rates. | $2.00 | $6.00 | 500K |
| Grok 4.5 Previous xAI frontier model for coding, agentic tasks, and knowledge work. xAI does not publish a separate maximum-output limit; prompts at or above 200K tokens use long-context pricing. | $2.00 | $6.00 | 500K |
| Grok 4.3 Lower-cost general-purpose xAI API model with configurable reasoning. xAI does not publish a separate maximum-output limit; requests above 200K tokens use long-context pricing. | $1.25 | $2.50 | 1.0M |
| Grok 4.20 Reasoning Pinned Grok 4.20 reasoning endpoint. xAI does not publish a separate maximum-output limit; requests above 200K tokens use long-context pricing. | $1.25 | $2.50 | 1.0M |
| Grok 4.20 Non-Reasoning Pinned Grok 4.20 endpoint without reasoning tokens. xAI does not publish a separate maximum-output limit; requests above 200K tokens use long-context pricing. | $1.25 | $2.50 | 1.0M |
| Grok 4.20 Multi-Agent Beta multi-agent endpoint for parallel research with 4-agent and 16-agent configurations. xAI does not publish a separate maximum-output limit. | $1.25 | $2.50 | 1.0M |
| Grok Build 0.1 Public-beta xAI model optimized for coding and software-building workflows. Requests above 200K tokens use long-context pricing; xAI does not publish a separate maximum-output limit. | $1.00 | $2.00 | 256K |
| Model | Input / 1M | Output / 1M | Context |
|---|---|---|---|
| Muse Wonder 1.0 Meta's most capable Muse model (preview), aimed at complex multi-file software engineering and long-horizon agent work. Accepts text, image, video, audio and PDF input. Standard mode only, with no long-context premium. Not available in South Korea. | $4.00 | $20.00 | 1.0M |
| Muse Spark 1.3 Meta's recommended Muse model for agentic and coding work at scale. Accepts text, image, video and PDF input (audio is not fully supported on 1.3). Muse Spark 1.2 and 1.1 remain available at the same price. | $1.25 | $4.25 | 1.0M |
| Muse Ember 1.0 Smallest and most efficient Muse model, for high-volume sub-agents, tool calls, routing and extraction. Requests above 200K input tokens bill entirely at the higher rates. Standard mode only. | $0.1 | $0.5 | 1.0M |
| Model | Input / 1M | Output / 1M | Context |
|---|---|---|---|
| Mistral Large 4 Open-weight multimodal Mixture-of-Experts flagship (52B active / 1.05T total parameters) with a 1M-token context. Mistral's recommended replacement for Large 3 and Medium 3.5. | $0.7 | $2.10 | 1.0M |
| Codestral 25.08 Mistral's low-latency code model for completion and fill-in-the-middle (FIM), with predicted outputs. Alias codestral-latest. | $0.3 | $0.9 | 128K |
| Ministral 3 14B Largest Ministral 3 model (Apache 2.0, open weights). Mistral says it performs comparably to Mistral Small 3.2 24B. Alias ministral-14b-latest. | $0.2 | $0.2 | 256K |
| Ministral 3 8B Efficient Ministral 3 edge model (Apache 2.0, open weights). Replaced ministral-8b-2410, retired 2025-12-31. Alias ministral-8b-latest. | $0.15 | $0.15 | 256K |
| Ministral 3 3B Smallest Ministral 3 edge model (Apache 2.0, open weights). Replaced ministral-3b-2410, retired 2025-12-31. Alias ministral-3b-latest. | $0.1 | $0.1 | 256K |
| Mistral Small 4 Current low-cost Mistral model for high-volume production routing. Mistral lists a 90% discount for cached input tokens. | $0.15 | $0.6 | 256K |
| Model | Input / 1M | Output / 1M | Context |
|---|---|---|---|
| DeepSeek V4.1 Flash Current DeepSeek Flash endpoint (DeepSeek-V4.1-Flash-1017, updated 2026-10-17 at unchanged prices) with native vision. Retired deepseek-v4-flash aliases temporarily route here. Peak hours are 01:00-04:00 and 06:00-10:00 UTC on weekdays; all other hours are 50% lower. | $0.3 | $1.20 | 1.0M |
| DeepSeek V4 Pro Serves DeepSeek-V4-Pro-0813. DeepSeek continued this endpoint after 2026-09-14 with unchanged billing. Peak hours are 01:00-04:00 and 06:00-10:00 UTC on weekdays; off-peak rates are 50% lower. | $1.32 | $3.96 | 1.0M |
| Model | Input / 1M | Output / 1M | Context |
|---|---|---|---|
| Xiaomi MiMo-V2.6-Pro Current full-modal MiMo flagship. Uses the former V2.5-Pro real-time price; Batch costs $0.2175/M cache-miss input and $0.435/M output. Cache writes are free for a limited time. | $0.435 | $0.87 | 1.0M |
| Xiaomi MiMo-V2.6-Pro-UltraSpeed Latency-focused V2.6-Pro endpoint with up to 20x output speed according to Xiaomi. Token rates are 10x Pro; Batch is unsupported. Contact Xiaomi for custom rate limits. | $4.35 | $8.70 | 1.0M |
| Xiaomi MiMo-V2.6-Flash Current full-modal high-volume MiMo model. Uses the former V2.5 real-time price; Batch costs $0.07/M cache-miss input and $0.14/M output. Cache writes are free for a limited time. | $0.14 | $0.28 | 1.0M |
| Model | Input / 1M | Output / 1M | Context |
|---|---|---|---|
| MiniMax M3 Latest MiniMax frontier coding and agent model at permanent 50%-off standard rates. Priority costs 1.5x Standard. The current official table does not publish M3 cache-write pricing. Supports Adaptive Thinking, tool use, and native image/video input. Inputs above 512K tokens use long-context pricing and currently require limited-access availability. | $0.3 | $1.20 | 1.0M |
| MiniMax M3.1 Flash Low-cost MiniMax model released 2026-10-19, half the price of MiniMax M3 on input and output. | $0.15 | $0.6 | 1.0M |
| Model | Input / 1M | Output / 1M | Context |
|---|---|---|---|
| GLM-5.3 Current GLM flagship for complex coding and long-horizon agents. Reasoning is always enabled with low, high, and max effort levels; cached-input storage is free for a limited time. | $1.40 | $4.40 | 1.0M |
| GLM-5.3 Flash Lowest-cost GLM-5.3 route. Cached-input storage is free for a limited time. | $0.15 | $0.5 | 1.0M |
| GLM-5.3 FlashX Faster GLM-5.3 API route. Cached-input storage is free for a limited time. | $0.37 | $1.25 | 1.0M |
| GLM-5.2 Previous GLM flagship. GLM-5.3 now has pay-as-you-go API pricing at the same token rates. | $1.40 | $4.40 | 1.0M |
| GLM-5.1 Previous GLM flagship for long-horizon coding and agent tasks. It remains available, while GLM-5.2 adds a 1M-token context window. | $1.40 | $4.40 | 200K |
| GLM-5 Previous-generation GLM flagship for agentic engineering (GLM-5.3 is current). Cached-input storage is free for a limited time. China pricing is tiered: ¥4/¥18 below 32K input, ¥6/¥22 at 32K and above. | $1.00 | $3.20 | 200K |
| GLM-4.7-Flash Free 30B-class version of GLM-4.7 on both Z.AI and bigmodel.cn; Z.AI lists it as completely free with no end date. | $0.00 | $0.00 | 200K |
| GLM-5 Turbo Current faster GLM model optimized for tool use and agent workflows. | $1.20 | $4.00 | 200K |
| GLM-5V Turbo Current multimodal GLM agent model for text, image, and video input. | $1.20 | $4.00 | 200K |
| Model | Input / 1M | Output / 1M | Context |
|---|---|---|---|
| Kimi K3 Kimi K3 global USD API rates. Automatic prefix caching uses a default 5-minute TTL with $3/M cache writes; one-hour writes cost $6/M. Cache hits cost $0.30/M and refresh the TTL without another write charge. Full model weights are available. | $3.00 | $15.00 | 1.0M |
| Kimi K3.1 Pro Newest Kimi flagship, cheaper than Kimi K3 ($3/$15). A separate Highspeed endpoint costs $4/$20 per 1M tokens. Cache hits cost $0.20/M; K3-series cache writes are billed by TTL. | $2.00 | $10.00 | 1.0M |
| Model | Input / 1M | Output / 1M | Context |
|---|---|---|---|
| Qwen4 Max Qwen4 flagship for complex reasoning and coding. International (Singapore) list price in USD, one tier up to 1M input tokens, same rate in thinking and non-thinking modes. Implicit and explicit cache hits cost $0.075/M; explicit cache writes cost $1.875/M (5-minute TTL) or $3/M (1-hour TTL). China (Beijing) and Global deployments list $1.413/$4.24 with $0.071/M cache hits. New Singapore users get 1M free tokens for 90 days. | $1.50 | $4.50 | 1.0M |
| Qwen3.8 Max Current Qwen flagship. China/global deployment price in CNY; Batch File input/output costs ¥6/¥18 per million tokens. Implicit cache hits cost ¥1.5/M and explicit cache hits ¥1/M, an exception to the generic cache ratios. | ¥12.00 | ¥36.00 | 1.0M |
| Qwen3.8 Flash Current lower-cost Qwen3.8 multimodal route (text, image and video input). China/global deployment price in CNY with no input-length tiers. Cache hits cost ¥0.1/M and explicit cache writes ¥1.25/M. | ¥0.8 | ¥2.70 | 1.0M |
| Qwen3.7 Flash Qwen3.7 native vision-language Flash model for multimodal agents and coding. International (Singapore) price is tiered by input length: ≤32K $0.03/$0.13, 32K–256K $0.10/$0.40, 256K–1M $0.20/$0.80 per 1M tokens; thinking and non-thinking modes cost the same. Cache hits shown are implicit-cache rates. Qwen3.8 Flash is the newer Flash model. | $0.03 | $0.13 | 1.0M |
| Qwen Plus Qwen3 hybrid thinking/non-thinking general model (alias of qwen-plus-2025-12-01). Prices shown are non-thinking; thinking-mode output costs $4/M (≤256K input) or $12/M above. Batch calls are half price. Cache hits shown are implicit-cache rates. | $0.4 | $1.20 | 1.0M |
| Qwen Flash Low-cost Qwen3 hybrid-thinking Flash model (alias of qwen-flash-2025-07-28). Cache hits shown are implicit-cache rates. | $0.05 | $0.4 | 1.0M |
| Qwen3.7 Plus Qwen3.7 Plus promotional price in China (20% off), listed in CNY. List price is ¥2/¥8 per million tokens below 256K and ¥6/¥24 above 256K. Cache list prices: hits ¥0.4/M and writes ¥2.5/M below 256K, ¥1.2/M and ¥7.5/M above. Alibaba lists no promotion end date. | ¥1.60 | ¥6.40 | 1.0M |
| Qwen3.7 Max Previous Qwen3.7 Max alias (text-only; image input arrived with the qwen3.7-max-2026-06-08 snapshot). Official China/global deployment list price is ¥12/M input and ¥36/M output; implicit cache hits ¥2.4/M, explicit cache hits ¥1.2/M. | ¥12.00 | ¥36.00 | 1.0M |
| Model | Input / 1M | Output / 1M | Context |
|---|---|---|---|
| Tencent Hy4 Tencent flagship Hunyuan model on TokenHub with deep reasoning, image input, function calling and caching. 1M context with up to 960K input and 64K output. USD list price from the international TokenHub price list. Replaces Hy4 preview, which Tencent decommissions on 2026-12-07. | $0.834 | $2.50 | 1.0M |
| Tencent Hy4 Flash Low-cost Hy4 model on TokenHub with the same 1M context, 64K output, deep reasoning, image input and function calling as Hy4, at about one twelfth of its list price. USD list price from the international TokenHub price list. | $0.07 | $0.281 | 1.0M |
| Model | Input / 1M | Output / 1M | Context |
|---|---|---|---|
| Command A+ Official docs now describe free API access within rate limits and Model Vault production deployment billed per instance-hour. The historical token rates above are unverified and excluded from the USD calculator. See https://docs.cohere.com/docs/command-a-plus and https://docs.cohere.com/docs/model-vault/standard/pricing. Current token pricing unverified | — | — | 128K |
| Model | Input / 1M | Output / 1M | Context |
|---|---|---|---|
| Jamba Large 1.7 Current Jamba Large API alias with a hybrid SSM-Transformer architecture and 256K context. The stable site slug is retained for existing links. | $2.00 | $8.00 | 256K |
| Jamba Mini 2 Current Jamba Mini API alias with 256K context. The stable site slug is retained for existing links. | $0.2 | $0.4 | 256K |
| Model | Input / 1M | Output / 1M | Context |
|---|---|---|---|
| Amazon Nova 2 Lite Current Amazon Nova 2 model on Bedrock for multimodal reasoning, tool use, and long-context workloads. | $0.3 | $2.50 | 1.0M |
| Amazon Nova Premier Most capable Amazon Nova 1 model and a teacher model for distillation, via Amazon Bedrock (us-east-1 on-demand; cross-region profile us.amazon.nova-premier-v1:0). Batch is 50% off ($1.25/$6.25). | $2.50 | $12.50 | 1.0M |
| Amazon Nova Pro Via AWS Bedrock. 300K context. | $0.8 | $3.20 | 300K |
| Amazon Nova Lite | $0.06 | $0.24 | 300K |
| Amazon Nova Micro Text-only. Lowest cost option on Bedrock. | $0.035 | $0.14 | 128K |