/ rankings

模型榜

围绕低价、中文、编码、视觉、Agent 和 OpenRouter 替代等决策场景打造的短名单,并附带来源标签和实际成本治理语境。

Blended price

Best cheap LLM API models for cost-sensitive products

Compare low-cost LLM API models by input price, output price, context length, capability, source, and production fit.

打开榜单
1
Doubao Seed 2.0 Mini

Chinese Q&A, low-cost general chat, multimodal understanding

VolcengineBlended price: CNY 2.200/1M
2
DeepSeek: DeepSeek V4 Flash

low-cost Chinese tasks, long-context summary, batch code assistance

DeepSeekBlended price: USD 0.336/1M
3
Mistral: Mistral Small 3.2 24B

translation, classification, short-form summarization

Mistral AIBlended price: USD 0.400/1M
4
OpenAI: GPT-4o-mini

low-cost chat, image understanding, classification

OpenRouterBlended price: USD 0.750/1M
5
Meta: Llama 4 Maverick

open-model workflows, cost-sensitive long context, classification

MetaBlended price: USD 0.750/1M
6
Google: Gemini 2.5 Flash

long-document summarization, image Q&A, fast multimodal routing

GoogleBlended price: USD 2.800/1M

Fit score

Best Chinese LLM API models for developer teams

Compare Chinese-language LLM API candidates across domestic and global providers, including pricing, context, latency estimates, and best use cases.

打开榜单
1
DeepSeek: R1

Chinese reasoning, math, analysis

DeepSeekFit score: 89/100
2
Doubao Seed 2.0 Mini

Chinese Q&A, low-cost general chat, multimodal understanding

VolcengineFit score: 88/100
3
DeepSeek: DeepSeek V4 Flash

low-cost Chinese tasks, long-context summary, batch code assistance

DeepSeekFit score: 88/100
4
Qwen: Qwen3 Coder Plus

Chinese engineering workflows, code generation, codebase Q&A

Alibaba Cloud / QwenFit score: 87/100
5
Qwen: Qwen3 Max

Chinese agent workflows, business analysis, structured output

Alibaba Cloud / QwenFit score: 86/100
6
MoonshotAI: Kimi K2.6

long Chinese documents, contract review, knowledge-base Q&A

Moonshot AIFit score: 84/100

Fit score

Best coding model APIs for agents and code review

Compare coding-oriented model APIs by context length, tool support, JSON output, latency estimate, price, and recommended production role.

打开榜单
1
Anthropic: Claude Opus 4.7

frontier reasoning, large codebase review, strategy analysis

AnthropicFit score: 96/100
2
Anthropic: Claude Sonnet 4.5

coding agents, code review, complex writing

AnthropicFit score: 93/100
3
DeepSeek: R1

Chinese reasoning, math, analysis

DeepSeekFit score: 89/100
4
DeepSeek: DeepSeek V4 Flash

low-cost Chinese tasks, long-context summary, batch code assistance

DeepSeekFit score: 88/100
5
Qwen: Qwen3 Coder Plus

Chinese engineering workflows, code generation, codebase Q&A

Alibaba Cloud / QwenFit score: 87/100

Fit score

Best vision model APIs for image understanding

Compare vision-capable model APIs for image understanding, document screenshots, multimodal support workflows, and cost-sensitive routing.

打开榜单
1
Anthropic: Claude Opus 4.7

frontier reasoning, large codebase review, strategy analysis

AnthropicFit score: 96/100
2
Anthropic: Claude Sonnet 4.5

coding agents, code review, complex writing

AnthropicFit score: 93/100
3
Google: Gemini 2.5 Pro

long-context analysis, vision workflows, scientific reasoning

GoogleFit score: 91/100
4
Doubao Seed 2.0 Mini

Chinese Q&A, low-cost general chat, multimodal understanding

VolcengineFit score: 88/100
5
Google: Gemini 2.5 Flash

long-document summarization, image Q&A, fast multimodal routing

GoogleFit score: 86/100
6
OpenAI: GPT-4o-mini

low-cost chat, image understanding, classification

OpenRouterFit score: 84/100

Catalog activity

OpenRouter alternatives for teams that need cost control

Compare OpenRouter-style multi-model access with cost governance, domestic provider coverage, BYOK, budget controls, and team usage reporting.

打开榜单
1
OpenAI: GPT-4o-mini

low-cost chat, image understanding, classification

OpenRouterCatalog activity: 93/100
2
Anthropic: Claude Sonnet 4.5

coding agents, code review, complex writing

AnthropicCatalog activity: 92/100
3
Anthropic: Claude Opus 4.7

frontier reasoning, large codebase review, strategy analysis

AnthropicCatalog activity: 90/100
4
DeepSeek: R1

Chinese reasoning, math, analysis

DeepSeekCatalog activity: 89/100
5
Google: Gemini 2.5 Pro

long-context analysis, vision workflows, scientific reasoning

GoogleCatalog activity: 88/100
6
Google: Gemini 2.5 Flash

long-document summarization, image Q&A, fast multimodal routing

GoogleCatalog activity: 86/100

Context

Best long-context model APIs for large documents

Compare long-context model APIs by context window, price, model source, and recommended document-heavy use cases.

打开榜单
1
Google: Gemini 2.5 Pro

long-context analysis, vision workflows, scientific reasoning

GoogleContext: 1049k tokens
2
DeepSeek: DeepSeek V4 Flash

low-cost Chinese tasks, long-context summary, batch code assistance

DeepSeekContext: 1049k tokens
3
Google: Gemini 2.5 Flash

long-document summarization, image Q&A, fast multimodal routing

GoogleContext: 1049k tokens
4
Meta: Llama 4 Maverick

open-model workflows, cost-sensitive long context, classification

MetaContext: 1049k tokens
5
Anthropic: Claude Opus 4.7

frontier reasoning, large codebase review, strategy analysis

AnthropicContext: 1000k tokens
6
Anthropic: Claude Sonnet 4.5

coding agents, code review, complex writing

AnthropicContext: 1000k tokens

Fit score

Best agent model APIs for tool-calling workflows

Compare model APIs for agent workflows that need tool calling, JSON mode, long context, and budget policies.

打开榜单
1
Anthropic: Claude Opus 4.7

frontier reasoning, large codebase review, strategy analysis

AnthropicFit score: 96/100
2
Anthropic: Claude Sonnet 4.5

coding agents, code review, complex writing

AnthropicFit score: 93/100
3
Google: Gemini 2.5 Pro

long-context analysis, vision workflows, scientific reasoning

GoogleFit score: 91/100
4
Qwen: Qwen3 Coder Plus

Chinese engineering workflows, code generation, codebase Q&A

Alibaba Cloud / QwenFit score: 87/100
5
Qwen: Qwen3 Max

Chinese agent workflows, business analysis, structured output

Alibaba Cloud / QwenFit score: 86/100
6
OpenAI: GPT-4o-mini

low-cost chat, image understanding, classification

OpenRouterFit score: 84/100