Compare AI Models
Filter and sort 44 AI models across 11 providers side-by-side - pricing, context window, type, and what each is best for.
Covers OpenAI, Anthropic, Google, Meta, DeepSeek, Microsoft, Amazon, Mistral, xAI, Cohere, and Qwen. Via-API pricing per 1M tokens as of Sep 2026. Always verify current pricing at the official provider pricing page before production use. Subscription plan pricing differs from API pricing.
Comparison Table
Showing 44 of 44 models
| Model | Provider | Type | Modalities | Params | Context ↕ | Input /1M ↑ | Output /1M ↕ | Tier | OW | Best for |
|---|---|---|---|---|---|---|---|---|---|---|
| Gemma 4 1B | General | TxtImgAud | 1B | 128K | $0.010 | $0.020 | Budget | ✓ | Extreme edge, mobile inference | |
| Llama 3.2 1B | Meta | General | Txt | 1B | 128K | $0.010 | $0.010 | Budget | ✓ | Smallest open-weight, edge & mobile |
| Mistral Nemo | Mistral | General | Txt | 12B | 128K | $0.020 | $0.040 | Budget | ✓ | Apache 2.0, ultra-budget open-weight |
| Gemma 4 4B | General | TxtImgAud | 4B | 128K | $0.030 | $0.060 | Budget | ✓ | On-device and edge deployments | |
| Phi-4-mini | Microsoft | General | Txt | 3.8B | 16K | $0.030 | $0.070 | Budget | ✓ | 3.8B ultra-small, on-device deployment |
| Nova Micro | Amazon | General | Txt | - | 128K | $0.035 | $0.14 | Budget | - | Text-only, lowest latency on AWS |
| Gemma 4 12B | General | TxtImgAud | 12B | 128K | $0.040 | $0.13 | Budget | ✓ | Efficient open-weight text, vision & audio | |
| Llama 3.2 11B Vision | Meta | General | TxtImg | 11B | 128K | $0.050 | $0.050 | Budget | ✓ | Open-weight vision & image understanding |
| Qwen3 8B | Qwen | General | Txt | 8B | 128K | $0.050 | $0.20 | Budget | ✓ | Small open-weight, local inference |
| Nova Lite | Amazon | General | TxtImgVid | - | 300K | $0.060 | $0.24 | Budget | - | Fast multimodal, very low cost on AWS |
| Phi-4 | Microsoft | General | Txt | 14B | 16K | $0.065 | $0.14 | Budget | ✓ | 14B - outperforms many larger text models |
| Phi-4-multimodal | Microsoft | General | TxtImgAud | 5.6B | 128K | $0.070 | $0.14 | Budget | ✓ | Audio + image + text, small multimodal model |
| Gemma 4 27B | General | TxtImgAud | 27B | 256K | $0.080 | $0.16 | Budget | ✓ | Open-weight multimodal + audio, self-hostable | |
| Phi-4 Mini Reasoning | Microsoft | Reasoning | Txt | 3.8B | 131K | $0.080 | $0.32 | Budget | ✓ | Small open-weight reasoning, 131K context |
| GPT-6 Luna | OpenAI | General | TxtImg | - | 1.05M | $0.10 | $0.50 | Budget | - | Cheapest GPT-6 generation model, high-volume tasks |
| Llama 4 Scout | Meta | General | TxtImg | 109B (MoE) | 10M | $0.10 | $0.35 | Budget | ✓ | Ultra-long 10M context window, multimodal |
| Llama 3.3 70B | Meta | General | Txt | 70B | 128K | $0.10 | $0.32 | Budget | ✓ | Open-weight text workhorse, widely deployed |
| Mistral Small 3.1 | Mistral | General | TxtImg | 24B | 128K | $0.10 | $0.30 | Budget | - | SOTA small model with vision, multilingual |
| DeepSeek V4 Flash | DeepSeek | General | Txt | - | 163K | $0.14 | $0.28 | Budget | ✓ | Ultra-low cost open-weight inference |
| Command R | Cohere | General | Txt | 35B | 128K | $0.15 | $0.60 | Budget | - | Efficient RAG for lighter workloads |
| Qwen3 32B | Qwen | General | Txt | 32B | 128K | $0.15 | $0.60 | Budget | ✓ | Strong open-weight, self-hostable |
| Llama 4 Maverick | Meta | General | TxtImg | 400B (MoE) | 1M | $0.20 | $0.85 | Budget | ✓ | Open-weight multimodal quality at low cost |
| Qwen3 Coder | Qwen | General | Txt | 480B (MoE) | 128K | $0.22 | $0.90 | Budget | ✓ | Alibaba coding specialist, open-weight |
| DeepSeek V4 Pro | DeepSeek | General | Txt | 1.6T (MoE) | 163K | $0.27 | $1.10 | Budget | ✓ | MIT license, frontier open-weight |
| Gemini 3.5 Flash-Lite | General | TxtImg | - | 1M | $0.30 | $2.50 | Budget | - | Cheapest current-generation Flash tier | |
| Codestral | Mistral | General | Txt | 22B | 256K | $0.30 | $0.90 | Budget | - | Code specialist - 80+ programming languages |
| DeepSeek R1 | DeepSeek | Reasoning | Txt | 671B | 163K | $0.55 | $2.19 | Budget | ✓ | Open-weight reasoning, matches frontier reasoning models |
| Gemini 3.8 Flash | General | TxtImgVidAud | - | 1M | $0.75 | $3.75 | Budget | - | Latest Flash, GA Sept 2026 - intro pricing through Dec 2026 | |
| Nova Pro | Amazon | General | TxtImgVid | - | 300K | $0.80 | $3.20 | Budget | - | Multimodal agentic workflows on AWS |
| Claude Haiku 4.5 | Anthropic | General | TxtImgDoc | - | 200K | $1.00 | $5.00 | Budget | - | High-volume extraction & classification |
| Grok 4.3 | xAI | Both | TxtImg | - | 131K | $1.25 | $2.50 | Mid | - | Previous-gen Grok, still widely available and cheaper |
| Mistral Medium 3.5 | Mistral | General | Txt | - | 128K | $1.50 | $7.50 | Mid | - | Current flagship-tier, multilingual tasks |
| Grok 4.7 | xAI | Both | TxtImg | - | 131K | $1.60 | $4.80 | Mid | - | Latest Grok (Sept 2026), frontier reasoning, X/Twitter knowledge |
| GPT-6 Astra | OpenAI | Both | TxtImgDoc | - | 1.05M | $2.00 | $10.00 | Mid | - | Flagship agentic model - operates software through screens, fills forms, edits spreadsheets |
| GPT-5.6 Terra | OpenAI | General | TxtImgDoc | - | 1M | $2.00 | $12.00 | Mid | - | Previous-generation mid-tier, still broadly deployed |
| Claude Sonnet 5 | Anthropic | Both | TxtImgDoc | - | 200K | $2.00 | $10.00 | Mid | - | Most agentic Sonnet yet, everyday production balance |
| Gemini 3.1 Pro | Both | TxtImgVidAudDoc | - | 1M | $2.00 | $12.00 | Mid | - | Frontier tier, top coding benchmarks (rises to $4/$18 above 200K tokens) | |
| Pixtral Large | Mistral | General | TxtImg | - | 128K | $2.00 | $6.00 | Mid | - | Multimodal flagship - vision + text |
| Nova Premier | Amazon | Both | TxtImgVidAud | - | 1M | $2.50 | $12.50 | Mid | - | Amazon flagship, extended thinking, 1M context |
| Command A | Cohere | General | Txt | - | 256K | $2.50 | $10.00 | Mid | - | Enterprise agentic workflows |
| Command R+ | Cohere | General | Txt | 104B | 128K | $2.50 | $10.00 | Mid | - | Enterprise RAG, grounded tool use |
| Claude Opus 5.5 | Anthropic | Both | TxtImgDoc | - | 1M | $4.00 | $20.00 | Mid | - | Near-Fable quality at a fraction of the cost, agentic pipelines |
| Qwen3.8 Max | Qwen | Both | Txt | - | 1M | $4.00 | $12.00 | Mid | ✓ | Alibaba flagship (Sept 2026), extended 1M context |
| Claude Fable 5.1 | Anthropic | Both | TxtImgDoc | - | 1M | $10.00 | $50.00 | Frontier | - | Most capable Claude - coding and knowledge work, always-on adaptive thinking |
OW = Open-weight (publicly released model weights). Click column headers to sort. Prices approximate via API (May 2026); always verify at the official provider pricing page before production use.
Reading This Table
Budget (≤ $1/1M input)
Best for high-volume, cost-sensitive workloads: classification, extraction, summarization at scale.
Mid ($1–$5/1M input)
Production workhorses. Good quality/cost ratio for most use cases: coding, analysis, chat.
Frontier (> $5/1M input)
Reserve for hard reasoning, complex agentic tasks, or where output quality is the primary constraint.
General vs Reasoning vs Both
General = standard chat/completion. Reasoning = uses internal thinking steps (DeepSeek R1). Both = supports both modes - e.g., Sonnet 5 with extended thinking on/off.
Open-weight
Open-weight models have publicly released weights. You can self-host them or access via third-party APIs (Groq, Together, Fireworks). Prices shown are typical API rates - self-hosting costs vary by GPU.