KotaML Logo
KotaML

LLM Token Pricing

(Pricing data as of December 2025)

Token pricing data from major LLM providers, grouped into Frontier, Mid, and OSS tiers. This page presents both median pricing by tier and detailed model-level input, output, and context window data.

Median Token Prices by Tier

Input ($ / 1M tokens)
Output ($ / 1M tokens)

Detailed Model Pricing

Model
Frontier HighAnthropicClaude Opus 4.5$5.00$25.00200,000
Frontier HighTogether.AICogito v2 preview – 405B$3.50$3.5032,000
Frontier HighTogether.AICogito v2 preview – 671B MoE$1.25$1.25128,000
Frontier HighTogether.AIDeepSeek-R1$3.00$7.00128,000
Frontier HighFireworks.AIDeepSeek-R1$1.35$5.40128,000
Frontier HighGoogleGemini 3 Pro$2.00$12.001,000,000
Frontier HighOpenAIGPT-5 pro$15.00$120.00400,000
Frontier HighOpenAIGPT-5.1$1.25$10.00400,000
Frontier HighFireworks.AILlama 3.1 405B Instruct Turbo$3.00$3.00128,000
Frontier HighTogether.AILlama 3.1 405B Instruct Turbo$3.50$3.50128,000
Frontier HighFireworks.AIQwen 3 Coder 480B$0.45$1.80256,000
Frontier HighTogether.AIQwen3-Coder 480B A35B Instruct$2.00$2.00256,000
Frontier ValueAnthropicClaude Haiku 4.5$1.00$5.00200,000
Frontier ValueAnthropicClaude Sonnet 4.5$3.00$15.00200,000
Frontier ValueTogether.AICogito v2 preview – 109B MoE$0.18$0.5932,000
Frontier ValueTogether.AIDeepSeek-V3$1.25$1.25128,000
Frontier ValueFireworks.AIDeepSeek-V3$0.56$1.68128,000
Frontier ValueOpenAIGPT-5 mini$0.25$2.00400,000
Frontier ValueOpenAIGPT-5 nano$0.05$0.40400,000
Frontier ValueFireworks.AIMixtral 8x22B$1.20$1.2032,000
Frontier ValueTogether.AIQwen3 235B A22B Thinking 2507 FP8$0.65$3.00256,000
Mid HighFireworks.AI>16B parameters $0.90$0.90
Mid HighTogether.AICogito v2 preview – 70B$0.88$0.8832,000
Mid HighTogether.AIDeepSeek R1 Distilled Llama 70B$2.00$2.00128,000
Mid HighGoogleGemini 2.5 Flash$0.30$2.501,000,000
Mid HighFireworks.AIGLM-4.6$0.55$2.19128,000
Mid HighTogether.AIGLM-4.6$0.60$2.20256,000
Mid HighTogether.AIKimi K2 Instruct$1.00$3.00128,000
Mid HighTogether.AIKimi K2 Thinking$1.20$4.00256,000
Mid HighFireworks.AIKimi K2 Thinking$0.60$2.508,000
Mid HighTogether.AILlama 3.1 70B Instruct Turbo$0.88$0.88128,000
Mid HighTogether.AILlama 3.3 70B Instruct-Turbo$0.88$0.88128,000
Mid HighFireworks.AILlama 4 Maverick$0.22$0.881,000,000
Mid HighTogether.AILlama 4 Maverick$0.27$0.851,000,000
Mid HighTogether.AIQwen 2.5 72B$1.20$1.2032,000
Mid HighTogether.AIQwen 3 235B$0.15$1.50128,000
Mid HighFireworks.AIQwen 3 235B$0.22$0.88128,000
Mid HighTogether.AIQwen 3 Next 80B$0.15$1.50256,000
Mid HighTogether.AIQwen3 Next 80B A3B Instruct$0.15$1.50262,000
Mid HighTogether.AIQwen3 Next 80B A3B Thinking$0.15$1.50256,000
Mid HighTogether.AITyphoon 2 70B Instruct$0.88$0.88128,000
Mid ValueFireworks.AI4B - 16B parameters $0.20$0.20
Mid ValueTogether.AIDeepSeek R1 Distilled Qwen 14B$0.18$0.18128,000
Mid ValueTogether.AILlama 3.1 8B Instruct Turbo$0.18$0.18128,000
Mid ValueFireworks.AILlama 4 Scout$0.15$0.601,000,000
Mid ValueTogether.AILlama 4 Scout$0.18$0.591,000,000
Mid ValueTogether.AIMistral Small 3$0.80$0.8032,000
Mid ValueTogether.AIMixtral 8x7B$0.60$0.6032,000
Mid ValueFireworks.AIMixtral 8x7B$0.50$0.5032,000
Mid ValueTogether.AIQwen2.5 7B Instruct Turbo$0.30$0.30128,000
OSS HighTogether.AIGLM-4.5 Air$0.20$1.10128,000
OSS HighTogether.AILlama 3 70B Instruct Reference$0.88$0.888,000
OSS HighTogether.AILlama 3 8B Instruct$0.10$0.108,000
OSS HighTogether.AIMistral 7B Instruct$0.20$0.2032,000
OSS HighFireworks.AIQwen 3 30B$0.15$0.60128,000
OSS HighTogether.AIQwen QwQ-32B$1.20$1.20128,000
OSS HighTogether.AIQwen2.5 Coder 32B Instruct$0.80$0.8032,000
OSS ValueFireworks.AI<4B Params$0.10$0.10
OSS ValueTogether.AIgemma-3n-E4B-it$0.02$0.0432,000
OSS ValueTogether.AILlama 3.2 3B Instruct Turbo$0.06$0.06128,000
OSS ValueTogether.AIOpenAI gpt-oss-20b$0.05$0.20128,000
OSS ValueFireworks.AIOpenAI gpt-oss-20b$0.07$0.30128,000

1) Pricing data as of December 2025. Values were taken from official provider pricing pages, including input, output, and max context window. Cached input tokens were excluded due to inconsistent reporting. Models without public pricing or context data were removed.

2) Models shown are the representative models for each pricing category. Tiers reflect list pricing, model family, expected performance class, and provider positioning. Tier medians are computed across these representative models. For a detailed explanation of the tier classifications, see the Model Tier Framework note.

3) This dataset reflects public list prices only; enterprise and volume discounts are not included.