LLM API price trend by brand

One dot per model: its release date against its currentcheapest blended price per 1M tokens (log scale), colored by the lab that made it. The market's whole story is here — each generation ships cheaper, while flagships hold the top. Click a brand to isolate it; click a dot for its page.

$0.05$0.10$0.30$1$3$10$30$60Jul 20242025Jul 20252026Jul 2026$ / 1M tokens (blended, log)Qwen 2.5 72B (DeepInfra) — $0.76/1M · released 2024-09-19 · cheapest via DeepInfraQwen3 Next 80B A3B Instruct (Alibaba) — $0.878/1M · released 2024-12-01 · cheapest via AlibabaQwen3 30B A3B Instruct 2507 (SiliconFlow) — $0.39/1M · released 2025-04-28 · cheapest via SiliconFlowQwen3 14B (DeepInfra) — $0.36/1M · released 2025-04-28 · cheapest via DeepInfraQwen3 32B (DeepInfra) — $0.36/1M · released 2025-04-30 · cheapest via DeepInfraQwen3 235B A22B Instruct 2507 (GMICloud) — $0.438/1M · released 2025-07-21 · cheapest via GMICloudQwen3 Coder (DeepInfra) — $1.3/1M · released 2025-07-22 · cheapest via DeepInfraQwen3 235B A22B Thinking 2507 (Alibaba) — $2.53/1M · released 2025-07-25 · cheapest via AlibabaQwen3 Coder 30B A3B Instruct (Novita) — $0.34/1M · released 2025-07-31 · cheapest via Novita AIQwen3 Next 80B A3B Thinking (Google) — $1.35/1M · released 2025-09-10 · cheapest via GoogleQwen3 VL 235B A22B Instruct (DeepInfra) — $1.08/1M · released 2025-09-23 · cheapest via DeepInfraQwen3 VL 235B A22B Thinking (Alibaba) — $4.4/1M · released 2025-09-23 · cheapest via AlibabaQwen3 VL 30B A3B Instruct (Alibaba) — $0.65/1M · released 2025-10-05 · cheapest via AlibabaQwen3 VL 30B A3B Thinking (SiliconFlow) — $1.29/1M · released 2025-10-11 · cheapest via SiliconFlowQwen3 VL 8B Instruct (Alibaba) — $0.572/1M · released 2025-10-14 · cheapest via AlibabaQwen3 Coder Next (Parasail) — $0.92/1M · released 2026-02-03 · cheapest via ParasailQwen3.5 397B A17B (Alibaba) — $2.73/1M · released 2026-02-15 · cheapest via AlibabaQwen3.5-35B-A3B (DeepInfra) — $1.14/1M · released 2026-02-23 · cheapest via DeepInfraQwen3.5-27B (Alibaba) — $1.755/1M · released 2026-02-23 · cheapest via AlibabaQwen3.5-9B (SiliconFlow) — $0.25/1M · released 2026-02-23 · cheapest via SiliconFlowQwen3.5-122B-A10B (SiliconFlow) — $2.34/1M · released 2026-02-23 · cheapest via SiliconFlowQwen3.6 35B A3B (AkashML) — $1/1M · released 2026-04-17 · cheapest via AkashMLQwen3.6 27B (Chutes) — $2.3/1M · released 2026-04-22 · cheapest via ChutesQwen3.8 2.4T A95B (Novita) — $8/1M · released 2026-08-12 · cheapest via Novita AIQwen3.8 27B (Parasail) — $2.44/1M · released 2026-08-14 · cheapest via ParasailMixtral 8x22B Instruct (Mistral) — $8/1M · released 2024-04-17 · cheapest via Mistral AIMistral Nemo (DeepInfra) — $0.049/1M · released 2024-07-01 · cheapest via DeepInfraMistral Large 3 (Mistral) — $2/1M · released 2024-11-01 · cheapest via Mistral AISaba (Mistral) — $0.8/1M · released 2025-02-17 · cheapest via Mistral AIMistral Medium (Mistral) — $9/1M · released 2025-05-07 · cheapest via Mistral AIMistral Medium 3 (Mistral) — $2.4/1M · released 2025-05-07 · cheapest via Mistral AIMistral Small 3.2 24B (DeepInfra) — $0.275/1M · released 2025-06-20 · cheapest via DeepInfraCodestral 2508 (Mistral) — $1.2/1M · released 2025-07-30 · cheapest via Mistral AIMistral Medium 3.1 (Mistral) — $2.4/1M · released 2025-08-13 · cheapest via Mistral AIVoxtral Small 24B 2507 (Mistral) — $0.4/1M · released 2025-10-30 · cheapest via Mistral AIMinistral 3 14B 2512 (Mistral) — $0.4/1M · released 2025-12-02 · cheapest via Mistral AIMinistral 3 8B 2512 (Mistral) — $0.3/1M · released 2025-12-02 · cheapest via Mistral AIMinistral 3 3B 2512 (Mistral) — $0.2/1M · released 2025-12-02 · cheapest via Mistral AIMistral Small (Mistral) — $0.75/1M · released 2026-03-16 · cheapest via Mistral AIMistral Medium 3.5 (Mistral) — $9/1M · released 2026-04-30 · cheapest via Mistral AIDeepSeek Chat (DeepInfra) — $1.21/1M · released 2024-12-26 · cheapest via DeepInfraDeepSeek V3 (DeepInfra) — $1.21/1M · released 2024-12-26 · cheapest via DeepInfraDeepSeek V3 0324 (DeepInfra) — $1.14/1M · released 2025-03-24 · cheapest via DeepInfraDeepSeek R1 (DeepInfra) — $2.65/1M · released 2025-05-28 · cheapest via DeepInfraDeepSeek V3.1 (DeepInfra) — $1.2/1M · released 2025-08-21 · cheapest via DeepInfraDeepSeek V3.1 Terminus (AtlasCloud) — $1.25/1M · released 2025-09-22 · cheapest via AtlasCloudDeepSeek V3.2 Exp (SiliconFlow) — $0.68/1M · released 2025-09-29 · cheapest via SiliconFlowDeepSeek V3.2 (GMICloud) — $0.518/1M · released 2025-12-01 · cheapest via GMICloudDeepSeek V4 Pro 0813 (Alibaba) — $2.323/1M · released 2026-04-24 · cheapest via AlibabaDeepSeek V4 Flash 0731 (Baidu) — $0.15/1M · released 2026-04-24 · cheapest via BaiduDeepSeek V4 Flash Vision Exp (DeepInfra) — $0.862/1M · released 2026-08-21 · cheapest via DeepInfraGLM 4.5 Air (Novita) — $0.98/1M · released 2025-07-28 · cheapest via Novita AIGLM 4.5V (Novita) — $2.4/1M · released 2025-08-11 · cheapest via Novita AIGLM 4.6 (Venice) — $2.18/1M · released 2025-09-30 · cheapest via VeniceGLM 4.6V (Novita) — $1.2/1M · released 2025-12-08 · cheapest via Novita AIGLM 4.7 (DeepInfra) — $2.15/1M · released 2025-12-22 · cheapest via DeepInfraGLM 4.7 Flash (Venice) — $0.46/1M · released 2026-01-19 · cheapest via VeniceGLM 5 (GMICloud) — $2.52/1M · released 2026-02-12 · cheapest via GMICloudGLM 5.1 (Baidu) — $3.764/1M · released 2026-04-07 · cheapest via BaiduGLM 5.2 (Baidu) — $1.479/1M · released 2026-06-13 · cheapest via BaiduGLM 5.3 (Reka) — $4.65/1M · released 2026-08-14 · cheapest via RekaGLM 5.3 Flash (Relace) — $0.309/1M · released 2026-08-26 · cheapest via RelaceGPT-4o — $12.5/1M · released 2024-05-13 · cheapest via OpenAIGPT-4o mini — $0.75/1M · released 2024-07-18 · cheapest via OpenAIGPT-4.1 mini — $2/1M · released 2025-04-14 · cheapest via OpenAIgpt-oss-20b (AkashML) — $0.12/1M · released 2025-08-05 · cheapest via AkashMLgpt-oss-120b (AkashML) — $0.2/1M · released 2025-08-05 · cheapest via AkashMLGPT-5.6 Sol (OpenAI) — $6/1M · released 2026-07-09 · cheapest via OpenAIGPT-5.6 Terra (OpenAI) — $7/1M · released 2026-07-09 · cheapest via OpenAIGPT-5.6 Luna (OpenAI) — $0.7/1M · released 2026-07-09 · cheapest via OpenAIGPT-6 Astra (OpenAI) — $60/1M · released 2026-09-03 · cheapest via OpenAIGemma 3 27B (DeepInfra) — $0.24/1M · released 2025-03-12 · cheapest via DeepInfraGemma 4 31B (CoreWeave) — $0.44/1M · released 2026-04-02 · cheapest via CoreWeaveGemma 4 26B A4B (Cloudflare) — $0.4/1M · released 2026-04-02 · cheapest via CloudflareGemini 3.6 Flash — $4.5/1M · released 2026-07-21 · cheapest via GoogleGemini 3.5 Flash Lite (Google) — $2.8/1M · released 2026-07-21 · cheapest via GoogleGemini 3.7 Flash (Google) — $4.5/1M · released 2026-08-13 · cheapest via GoogleGemini 3.8 Flash (Google) — $4.5/1M · released 2026-09-02 · cheapest via GoogleLlama 3.1 8B (DeepInfra) — $0.06/1M · released 2024-07-23 · cheapest via DeepInfraLlama 3.1 70B Instruct (DeepInfra) — $0.8/1M · released 2024-07-23 · cheapest via DeepInfraLlama 3.2 3B Instruct (Parasail) — $0.38/1M · released 2024-09-18 · cheapest via ParasailLlama 3.3 70B (DeepInfra) — $0.42/1M · released 2024-12-06 · cheapest via DeepInfraLlama 4 Scout (DeepInfra) — $0.4/1M · released 2025-04-05 · cheapest via DeepInfraLlama 4 Maverick (DigitalOcean) — $0.896/1M · released 2025-04-05 · cheapest via DigitalOceanMiniMax M1 (Minimax) — $2.6/1M · released 2025-06-17 · cheapest via MinimaxMiniMax M2 (Minimax) — $1.275/1M · released 2025-10-27 · cheapest via MinimaxMiniMax M2.1 (Novita) — $1.5/1M · released 2025-12-23 · cheapest via Novita AIMiniMax M2.5 (Venice) — $1.22/1M · released 2026-02-12 · cheapest via VeniceMiniMax M2.7 (Novita) — $1.35/1M · released 2026-03-18 · cheapest via Novita AIMiniMax M3 (CoreWeave) — $1.19/1M · released 2026-06-01 · cheapest via CoreWeaveKimi K2 Thinking (Google) — $3.1/1M · released 2025-11-06 · cheapest via GoogleKimi K2.5 (SiliconFlow) — $2.7/1M · released 2026-01-27 · cheapest via SiliconFlowKimi K2.6 (Baidu) — $2.856/1M · released 2026-04-21 · cheapest via BaiduKimi K2.7 Code (DeepInfra) — $4.08/1M · released 2026-06-12 · cheapest via DeepInfraKimi K3 (DeepInfra) — $17.1/1M · released 2026-07-16 · cheapest via DeepInfraClaude Fable 5 (Anthropic) — $60/1M · released 2026-06-09 · cheapest via AnthropicClaude Sonnet 5 (Anthropic) — $12/1M · released 2026-06-30 · cheapest via AnthropicClaude Opus 5 (Anthropic) — $30/1M · released 2026-07-24 · cheapest via AnthropicClaude Fable 5.1 (Anthropic) — $60/1M · released 2026-09-01 · cheapest via AnthropicNemotron 3 Nano 30B A3B (Crusoe) — $0.25/1M · released 2025-12-15 · cheapest via CrusoeNemotron 3 Super (DeepInfra) — $0.485/1M · released 2026-03-11 · cheapest via DeepInfraNemotron 3 Ultra (DeepInfra) — $2.7/1M · released 2026-06-04 · cheapest via DeepInfraNemotron 3.5 Lightning (DeepInfra) — $0.28/1M · released 2026-08-11 · cheapest via DeepInfraCommand R (Cohere) — $0.75/1M · released 2024-08-30 · cheapest via CohereCommand A (Cohere) — $12.5/1M · released 2025-03-13 · cheapest via CohereGrok 4.5 (xAI) — $8/1M · released 2026-07-08 · cheapest via xAIGrok 4.6 (xAI) — $8/1M · released 2026-08-12 · cheapest via xAIGranite 4.2 8B (CoreWeave) — $0.25/1M · released 2026-08-31 · cheapest via CoreWeave

108 models, verified 2026-09-05. Each dot = one base model at its cheapest live offer (blended = input + output per 1M tokens; $0 free tiers excluded). Release dates from the index referential. Prices change hands daily; this page rebuilds with them.

FAQ

Why do LLM API prices keep falling?

Each model generation ships cheaper at equal or better capability: competition between labs, cheaper inference hardware, and distillation of large models into small ones. On this chart the spread runs from $0.049/1M (Mistral Nemo (DeepInfra)) to $60/1M (GPT-6 Astra (OpenAI)) — three orders of magnitude, all currently sold.

Does a model's price change after release?

Rarely, and mostly downward: open-weight models get cheaper as hosts compete, and labs occasionally cut list prices. The dominant effect by far is generational — new models replace old ones at lower prices. This chart plots each model's current cheapest offer against its release date.