Directory
AI models: prices, context windows and grid references
286 models from 13 creators, 93 of them open-weight. Prices are per 1M tokens on each model’s cheapest available standard-tier offer, each one linked to its source and date; the newest was checked 12 Oct 2026. 180 have a standard-tier price and a context window, so they sit on the atlas.
Ask an AI about this pageGChatGPTCClaudePPerplexityGeGeminiXGrok
Cite
Cite this page
ModelsAtlas. “AI models: prices, context windows and grid references”. Retrieved 12 Oct 2026. https://modelsatlas.com/models/page/5
Showing 201–250 of 286 models · cheapest input first, models without a standard-tier price last
| Model | Grid | Context | Input / 1M | Output / 1M |
|---|---|---|---|---|
| gemma 4 31B it qat w4a16 ctGoogle · open weights | — | — | — | — |
| gemma 4 E2B itGoogle · open weights | — | — | — | — |
| gemma 4 E4B itGoogle · open weights | — | — | — | — |
| GPT-5 Codex (batch)OpenAI | — | 400K | — | — |
| Llama 3_1 Nemotron Ultra 253B v1NVIDIA · open weights | — | 131K | — | — |
| Llama 3_3 Nemotron Super 49B v1_5NVIDIA · open weights | — | 131K | — | — |
| Llama 3.1 405BMeta · open weights | — | — | — | — |
| Llama 3.1 405B InstructMeta · open weights | — | — | — | — |
| Llama 3.1 Nemotron 70B InstructNVIDIA · open weights | — | — | — | — |
| Llama 3.2 11B Vision InstructMeta · open weights | — | — | — | — |
| Llama 3.2 1BMeta · open weights | — | — | — | — |
| Llama Guard 3 8BMeta · open weights | — | — | — | — |
| mdeberta v3 baseMicrosoft · open weights | — | 512 | — | — |
| Meta Llama 3 70B InstructMeta · open weights | — | — | — | — |
| Meta Llama 3 8BMeta · open weights | — | — | — | — |
| Meta Llama 3 8B InstructMeta · open weights | — | — | — | — |
| Mistral 7B Instruct v0.1Mistral AI · open weights | — | 33K | — | — |
| Mistral 7B Instruct v0.2Mistral AI · open weights | — | 33K | — | — |
| Mistral 7B Instruct v0.3Mistral AI · open weights | — | 33K | — | — |
| Nemotron 3 Nano Omni (free)NVIDIA | — | 256K | — | — |
| Nemotron Nano 12B 2 VL (free)NVIDIA | — | 128K | — | — |
| North Mini Code (free)Cohere | — | 256K | — | — |
| NVIDIA Nemotron 3 Super 120B A12B NVFP4NVIDIA · open weights | — | 262K | — | — |
| NVIDIA Nemotron Nano 9B v2NVIDIA · open weights | — | 131K | — | — |
| phi 2Microsoft · open weights | — | 2K | — | — |
| Prompt Guard 86MMeta · open weights | — | — | — | — |
| Qwen 72BQwen (Alibaba) · open weights | — | 33K | — | — |
| Qwen2 1.5B InstructQwen (Alibaba) · open weights | — | 33K | — | — |
| Qwen2 VL 2B InstructQwen (Alibaba) · open weights | — | 33K | — | — |
| Qwen2.5 0.5BQwen (Alibaba) · open weights | — | 33K | — | — |
| Qwen2.5 0.5B InstructQwen (Alibaba) · open weights | — | 33K | — | — |
| Qwen2.5 1.5B InstructQwen (Alibaba) · open weights | — | 33K | — | — |
| Qwen2.5 14B InstructQwen (Alibaba) · open weights | — | 33K | — | — |
| Qwen2.5 32B InstructQwen (Alibaba) · open weights | — | 33K | — | — |
| Qwen2.5 3B InstructQwen (Alibaba) · open weights | — | 33K | — | — |
| Qwen2.5 7BQwen (Alibaba) · open weights | — | 131K | — | — |
| Qwen2.5 Coder 14B InstructQwen (Alibaba) · open weights | — | 33K | — | — |
| Qwen2.5 Coder 7B InstructQwen (Alibaba) · open weights | — | 33K | — | — |
| Qwen2.5 VL 32B InstructQwen (Alibaba) · open weights | — | 128K | — | — |
| Qwen2.5 VL 3B InstructQwen (Alibaba) · open weights | — | 128K | — | — |
| Qwen2.5 VL 7B InstructQwen (Alibaba) · open weights | — | 128K | — | — |
| Qwen3 0.6BQwen (Alibaba) · open weights | — | 41K | — | — |
| Qwen3 1.7BQwen (Alibaba) · open weights | — | 41K | — | — |
| Qwen3 4BQwen (Alibaba) · open weights | — | 41K | — | — |
| Qwen3 4B Instruct 2507Qwen (Alibaba) · open weights | — | 262K | — | — |
| Qwen3 8BQwen (Alibaba) | — | 131K | — | — |
| Qwen3 ASR 0.6BQwen (Alibaba) · open weights | — | — | — | — |
| Qwen3 ASR 1.7BQwen (Alibaba) · open weights | — | — | — | — |
| Qwen3 Coder PlusQwen (Alibaba) | — | 1M | — | — |
| Qwen3 Embedding 0.6BQwen (Alibaba) · open weights | — | 33K | — | — |
Standard-tier prices from the cheapest available offer. “—” means we hold no price, or no context window, for it yet.Suggest a correction
Browse by creator
About this directory
One row per model page. Size variants (mini, pro, flash) and serving tiers (batch, free, thinking) belong to the same model, so they appear as rows on its page, not as separate entries. A grid reference places a model by price and context; it is not a quality score. See our methodology.