So sanh cac model API LLM chi phi thap theo gia input, gia output, boi canh, capability, nguon va do phu hop voi production.
Viec chon API LLM gia re nen bat dau tu hinh dang workload, khong chi tu muc gia thap nhat dang hien thi. Doi voi classification, summarization, routing, support draft va batch transformation, mot model re hon co the giam chi phi hang thang ma khong can thay doi giao dien ung dung. Doi voi final answer, reasoning phuc tap hoac coding agent, doi ngu nen so sanh model gia re voi mot fallback manh hon. NextModel tap hop gia, boi canh, capability, nguon provider va vi du ma nguon trong mot noi truoc khi len production.
Co so nguon: Catalog duoc chon loc cua NextModel, gia cong khai tu provider va metadata OpenRouter khi co san.
Blended price
Bat dau voi danh sach rut gon, thu prompt thuc te va so sanh chi phi hang thang truoc khi routing production.
DeepSeek V4 Flash is an efficiency-optimized Mixture-of-Experts model from DeepSeek with 284B total parameters and 13B activated parameters, supporting a 1M-token context window. It is designed for fast inference and...
Mistral-Small-3.2-24B-Instruct-2506 is an updated 24B parameter model from Mistral optimized for instruction following, repetition reduction, and improved function calling. Compared to the 3.1 release, version 3.2 significantly improves accuracy on...
GPT-4o mini is OpenAI's newest model after [GPT-4 Omni](/models/openai/gpt-4o), supporting both text and image inputs with text outputs. As their most advanced small model, it is many multiples more affordable...
Llama 4 Maverick 17B Instruct (128E) is a high-capacity multimodal language model from Meta, built on a mixture-of-experts (MoE) architecture with 128 experts and 17 billion active parameters per forward...
Bang so sanh
Dung giao dien nay de thu hep shortlist production, xay dung chinh sach fallback hoac so sanh kinh te model.
| Model | Provider | Input | Output | Context | Capabilities | Best for | Latency | Status | Source |
|---|---|---|---|---|---|---|---|---|---|
| DeepSeek: DeepSeek V4 Flashdeepseek/deepseek-v4-flash | DeepSeek | $0.112 / 1M tokens | $0.224 / 1M tokens | 1M | Tool callingJSON modeLong contextReasoning | low-cost Chinese tasks, long-context summary | 800-2600ms | Catalog | OpenRouter if available |
| Mistral: Mistral Small 3.2 24Bmistralai/mistral-small-3.2-24b-instruct | Mistral AI | $0.1 / 1M tokens | $0.3 / 1M tokens | 128k | Tool callingJSON modeStreamingLow cost | translation, classification | 700-2300ms | Catalog | OpenRouter if available |
| OpenAI: GPT-4o-miniopenai/gpt-4o-mini | OpenRouter | $0.15 / 1M tokens | $0.6 / 1M tokens | 128k | Tool callingVisionJSON modeLong context | low-cost chat, image understanding | 800-2400ms | Catalog | OpenRouter if available |
| Meta: Llama 4 Maverickmeta-llama/llama-4-maverick | Meta | $0.15 / 1M tokens | $0.6 / 1M tokens | 1M | JSON modeLong contextStreamingLow cost | open-model workflows, cost-sensitive long context | 950-2800ms | Catalog | OpenRouter if available |
| Google: Gemini 2.5 Flashgoogle/gemini-2.5-flash | $0.3 / 1M tokens | $2.50 / 1M tokens | 1M | Tool callingVisionJSON modeLong context | long-document summarization, image Q&A | 900-2800ms | Catalog | OpenRouter if available | |
| MoonshotAI: Kimi K2.6moonshotai/kimi-k2.6 | Moonshot AI | $0.73 / 1M tokens | $3.49 / 1M tokens | 262.1k | JSON modeLong contextStreamingTool calling | long Chinese documents, contract review | 1400-4400ms | Catalog | OpenRouter if available |
FAQ
Dieu nay phu thuoc vao ty gia va do dai output. Doubao Seed 2.0 Mini van la lua chon production bang CNY co chi phi thap nhat trong catalog nay.
Khong. Model gia re phu hop voi cong viec lap lai va rui ro thap; doi voi final answer, reasoning phuc tap va coding agent, can so sanh chung voi cac model manh hon.