Comparisons
AI Model Comparison
21 large language models from 7 providers, compared on what each provider officially documents: context window, input types and list API prices. Every row links to its source and shows when we last verified it.
Showing 21 of 21 models
| Model | Provider | Context window | Inputs | Documented focus | API price / 1M tokens input · output (USD) |
Verified | Official source |
|---|---|---|---|---|---|---|---|
Claude 3.5 Haikuclaude-3-5-haiku-latest |
Anthropic | 200K tokens | text, image | fast, cheap | $0.8 · $4 | Docs for Claude 3.5 Haiku (opens in a new tab) | |
Claude Opus 4.1claude-opus-4-1 |
Anthropic | 200K tokens | text, image | deep reasoning, coding | $15 · $75 | Docs for Claude Opus 4.1 (opens in a new tab) | |
Claude Sonnet 4.5claude-sonnet-4-5 |
Anthropic | 200K tokens | text, image | coding, reasoning, long context | $3 · $15 | Docs for Claude Sonnet 4.5 (opens in a new tab) | |
DeepSeek R1 (reasoner)deepseek-reasoner |
DeepSeek | 64K tokens | text | reasoning | $0.55 · $2.19 | Docs for DeepSeek R1 (reasoner) (opens in a new tab) | |
DeepSeek V3deepseek-chat |
DeepSeek | 64K tokens | text | general, cheap | $0.27 · $1.1 | Docs for DeepSeek V3 (opens in a new tab) | |
Gemini 2.0 Flashgemini-2.0-flash |
1M tokens | text, image | fast, cheap | $0.1 · $0.4 | Docs for Gemini 2.0 Flash (opens in a new tab) | ||
Gemini 2.5 Flashgemini-2.5-flash |
1M tokens | text, image, audio | fast, multimodal | $0.3 · $2.5 | Docs for Gemini 2.5 Flash (opens in a new tab) | ||
Gemini 2.5 Progemini-2.5-pro |
1M tokens | text, image, audio | reasoning, long context | $1.25 · $10 | Docs for Gemini 2.5 Pro (opens in a new tab) | ||
Mistral Largemistral-large-latest |
Mistral AI | 128K tokens | text | general, multilingual | $2 · $6 | Docs for Mistral Large (opens in a new tab) | |
Mistral Smallmistral-small-latest |
Mistral AI | 128K tokens | text | fast, cheap | $0.1 · $0.3 | Docs for Mistral Small (opens in a new tab) | |
GPT-4.1gpt-4.1 |
OpenAI | 1M tokens | text, image | coding, long context | $2 · $8 | Docs for GPT-4.1 (opens in a new tab) | |
GPT-4.1 minigpt-4.1-mini |
OpenAI | 1M tokens | text, image | fast, cheap | $0.4 · $1.6 | Docs for GPT-4.1 mini (opens in a new tab) | |
GPT-4ogpt-4o |
OpenAI | 128K tokens | text, image, audio | general, multimodal | $2.5 · $10 | Docs for GPT-4o (opens in a new tab) | |
GPT-4o minigpt-4o-mini |
OpenAI | 128K tokens | text, image | fast, cheap | $0.15 · $0.6 | Docs for GPT-4o mini (opens in a new tab) | |
o3o3 |
OpenAI | 200K tokens | text, image | reasoning, analysis | $2 · $8 | Docs for o3 (opens in a new tab) | |
o4-minio4-mini |
OpenAI | 200K tokens | text, image | reasoning, fast | $1.1 · $4.4 | Docs for o4-mini (opens in a new tab) | |
DeepSeek V3 (via OpenRouter)deepseek/deepseek-chat |
OpenRouter | 64K tokens | text | general, cheap | Pricing varies by routing provider — check OpenRouter. | Docs for DeepSeek V3 (via OpenRouter) (opens in a new tab) | |
Llama 3.3 70B (Meta)meta-llama/llama-3.3-70b-instruct |
OpenRouter | 128K tokens | text | open weights | Pricing varies by routing provider — check OpenRouter. | Docs for Llama 3.3 70B (Meta) (opens in a new tab) | |
Qwen 2.5 72B (via OpenRouter)qwen/qwen-2.5-72b-instruct |
OpenRouter | 32K tokens | text | multilingual | Pricing varies by routing provider — check OpenRouter. | Docs for Qwen 2.5 72B (via OpenRouter) (opens in a new tab) | |
Grok 3grok-3 |
xAI | 131K tokens | text | general, reasoning | $3 · $15 | Docs for Grok 3 (opens in a new tab) | |
Grok 3 minigrok-3-mini |
xAI | 131K tokens | text | fast, reasoning | $0.3 · $0.5 | Docs for Grok 3 mini (opens in a new tab) |
How to read this table
- Source of truth: every value comes from the provider's own documentation, linked in the last column. The registry was last verified on Oct 5, 2026.
- Prices are list prices per million tokens for API use, in US dollars. Providers change prices, offer discounts and bill some features separately — confirm on the provider's pricing page before you budget.
- Documented focus summarises what the provider positions the model for (e.g. reasoning, speed). It is not a benchmark score — we don't publish rankings or invented benchmarks.
- Missing a newer model? Providers release models frequently. We add a model only after checking its official documentation.