At a glance
Newest release
GPT-6.1 Terra
OpenAI. Released Oct 26, 2026
Lowest list price
Ministral 3 8B
Mistral AI. $0.15 in, $0.15 out per 1M tokens
Lowest price with open weights
Ministral 3 8B
Mistral AI. $0.15 in, $0.15 out per 1M tokens
Lowest price that reads audio and video
Qwen3.8 Omni Flash
Alibaba. $0.15 in, $0.47 out per 1M tokens
Compare AI models
31 models
| Model | Max output | Reads | Can use | Check | |||
|---|---|---|---|---|---|---|---|
| GPT-6.1 TerraOpenAI |
Oct 26, 2026 | 1.05M | 128K | TextImagesPDFs | ToolsReasoning |
$1 in $5 out |
Verified |
 Kimi K3.1 ProMoonshot AI |
Oct 23, 2026 | 1.05M | — | TextImagesVideo | ToolsReasoning |
$2 in $10 out |
Verified |
 DeepSeek V4 FlashDeepSeek |
Oct 17, 2026 | — | — | TextImages | ToolsReasoningOpen weights | See company page | Listed by company |
 Gemini 4 ArgonGoogle |
Oct 13, 2026 | 1.05M | — | TextImagesPDFsAudioVideo | ToolsReasoning |
$2 in $10 out |
Verified |
 Muse WonderMeta |
Oct 12, 2026 | 1.05M | — | TextImagesPDFsAudioVideo | ToolsReasoning |
$4 in $20 out |
Verified |
 Claude Haiku 5.5Anthropic |
Oct 7, 2026 | 1M | 128K | TextImagesPDFs | ToolsReasoning |
$0.10 in $0.50 out |
Verified |
| Mistral Large 4Mistral AI |
Oct 6, 2026 | — | — | TextImagesPDFs | ToolsReasoning | See company page | Listed by company |
| GPT-6.1 SolOpenAI |
Sep 29, 2026 | 1.05M | 128K | TextImagesPDFs | ToolsReasoning |
$2 in $10 out |
Verified |
| Claude Sonnet 5.5Anthropic |
Sep 28, 2026 | 1M | 128K | TextImagesPDFs | ToolsReasoning |
$2 in $10 out |
Verified |
| Claude Opus 5.5Anthropic |
Sep 22, 2026 | 1M | 128K | TextImagesPDFs | ToolsReasoning |
$4 in $20 out |
Verified |
| GPT-6 LunaOpenAI |
Sep 22, 2026 | 1.05M | 128K | TextImagesPDFs | ToolsReasoning |
$0.10 in $0.50 out |
Verified |
 Grok 4.7xAI |
Sep 21, 2026 | 500K | 500K | TextImagesPDFs | ToolsReasoning |
$2 in $6 out |
Verified |
| Qwen3.8 Omni FlashAlibaba |
Sep 17, 2026 | 1M | 131K | TextImagesAudioVideo | ToolsReasoning |
$0.15 in $0.47 out |
Verified |
| GPT-6 AstraOpenAI |
Sep 4, 2026 | 1.05M | 128K | TextImagesPDFs | ToolsReasoning |
$10 in $50 out |
Verified |
| Gemini 3.8 FlashGoogle |
Sep 2, 2026 | — | — | TextImagesPDFsAudioVideo | ToolsReasoning | See company page | Listed by company |
| Muse Spark 1.3Meta |
Sep 2, 2026 | 1.05M | — | TextImagesPDFsAudioVideo | ToolsReasoning |
$1.25 in $4.25 out |
Verified |
| Claude Fable 5.1Anthropic |
Sep 1, 2026 | 1M | 128K | TextImagesPDFs | ToolsReasoning |
$10 in $50 out |
Verified |
| Qwen3.8 FlashAlibaba |
Aug 26, 2026 | 1M | 131K | TextImagesVideo | ToolsReasoning |
$0.15 in $0.47 out |
Verified |
| DeepSeek V4 ProDeepSeek |
Aug 13, 2026 | — | — | Text | ToolsReasoningOpen weights | See company page | Listed by company |
| Qwen3.8 MaxAlibaba |
Aug 3, 2026 | — | — | TextImagesPDFsVideo | ToolsReasoning | See company page | Listed by company |
| Gemini 3.5 Flash LiteGoogle |
Jul 21, 2026 | 1.05M | 66K | TextImagesPDFsAudioVideo | ToolsReasoning |
$0.30 in $2.50 out |
Verified |
| Kimi K2.7 CodeMoonshot AI |
Jun 12, 2026 | — | — | TextImagesVideo | ToolsReasoningOpen weights | See company page | Listed by company |
 North Mini CodeCohere |
Jun 9, 2026 | — | — | Text | ToolsReasoningOpen weights | See company page | Listed by company |
| Qwen3.7 PlusAlibaba |
Jun 2, 2026 | — | — | TextImagesVideo | ToolsReasoning | See company page | Listed by company |
| Command A PlusCohere |
May 20, 2026 | — | — | TextImages | ToolsReasoningOpen weights | See company page | Listed by company |
| Mistral Medium 3.5Mistral AI |
Apr 29, 2026 | — | — | TextImagesPDFs | ToolsReasoningOpen weights | See company page | Listed by company |
| GPT-5.5OpenAI |
Apr 23, 2026 | 1.05M | 128K | TextImagesPDFs | ToolsReasoning |
$5 in $30 out |
Sources agree |
| Grok Build 0.1xAI |
Apr 16, 2026 | 256K | 256K | TextImagesPDFs | ToolsReasoning |
$1 in $2 out |
Verified |
| Mistral Small 4Mistral AI |
Mar 16, 2026 | 262K | — | TextImagesPDFs | ToolsReasoningOpen weights |
$0.15 in $0.60 out |
Verified |
| Devstral 2Mistral AI |
Dec 9, 2025 | 262K | — | Text | ToolsOpen weights |
$0.40 in $2 out |
Verified |
| Ministral 3 8BMistral AI |
Dec 2, 2025 | 262K | — | TextImagesPDFs | ToolsOpen weights |
$0.15 in $0.15 out |
Verified |
No model matches both filters. Change one of them.
Prices are base list prices for the company's own API. Long prompts, batch work, caching and other platforms can change them, so confirm on the company's page before you budget. A dash or "See company page" means we could not confirm the figure, so we leave it out.
Recent changes
- Ministral 3 8B (Mistral AI): added to the table
- Ministral 3 3B (Mistral AI): added to the table
Which model for which job
Start from the job, then use the table to narrow the list. Test the finalists on your own examples before you commit.
Customer support chat
Look for tool use and a low price per token, because volume drives the bill.
AI Customer SupportDocuments and invoices
Look for PDF input and a large context window, so long files fit in one request.
AI Document ProcessingAgents that take actions
Look for tool use and reasoning, then test on real tasks with human review.
AI Agent DevelopmentKeeping data in-house
Look for open weights, which you can run on infrastructure you control.
Open-Source LLM DevelopmentSources and labels
Data from models.dev (MIT license), the OpenRouter model list and each company's official pages. Also available as JSON.
- Verified On the company's official page, with figures our sources agree on.
- Listed by company On the company's official page. See that page for current figures.
- Sources agree Our sources agree on the figures. Confirm on the company's page before you budget.
Product and company names and logos belong to their owners. Aranv is not affiliated with these companies. This table lists facts, not a ranking of quality.
AI model comparison FAQs
Which AI model is best for a business?
There is no single best model. It depends on the job, your volume and your data rules. Use the table to narrow the list by what the model can read and what it costs, then test the finalists on your own examples.
What is the difference between ChatGPT and GPT models?
ChatGPT is OpenAI's app, and GPT models are what you call through OpenAI's API when you build your own software. The same split applies to the Claude and Gemini apps and their API models, and limits and prices can differ between them.
What is a context window?
The amount of text a model can work with in one request, counted in tokens. A larger window lets a model read longer documents or conversations at once.
What do the prices in the table mean?
The base list price per one million tokens: "in" for the text you send and "out" for the text the model writes. Long prompts, batch work and caching can change the price, so confirm on the company's page.
Why do some models say "See company page"?
Because we could not confirm the figure. Rather than guess, we leave the number out and link to the company's pricing page.
What does "open weights" mean?
The company publishes the trained model so you can run it on infrastructure you control. Check the license, which sets out how you may use it.
Need help choosing and building?
Tell us the workflow and the data involved. We will help you test the right models and build the system around them.