> ## Documentation Index
> Fetch the complete documentation index at: /docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Model Comparison

> Compare AI models across all providers

## Overview

Compare 94 AI models across 14 providers to find the right one for your use case — by speed, cost, reasoning capability and context window.

## Quick Comparison by Use Case

Credits are charged per step: each model call in an AI teammate turn, or one chat reply. A model's figure is its provider
price for a typical step — about 31K prompt tokens, most of them cached, and 465 output tokens.

> [card] Best Overall Value
>
> **GPT-6 Luna** - 1 credit
>
> Reasoning, vision and a 1.05M context at the lowest price per step

> [card] Best Mid-Range
>
> **Gemini 3.8 Flash** - 3 credits
>
> Strong agentic coding with a 1M context

> [card] Best Reasoning
>
> **GPT-6.1 Sol** - 8 credits
>
> Near-Astra performance for complex work

> [card] Largest Context
>
> **Claude Sonnet 5.5** - 8 credits
>
> 1M tokens with adaptive thinking

> [card] Most Affordable
>
> **GPT-5 Nano** - 1 credit
>
> Lowest cost per step for simple, high-volume work

> [card] Best for Long Agentic Work
>
> **Claude Opus 5.5** - 15 credits
>
> Long-running coding and knowledge work

## All Models Comparison

### OpenAI Models

| Model | Model id | Credits | Context | Speed | Cost | Reasoning | Vision |
| - | - | - | - | - | - | - | - |
| **GPT-6 Astra** | `gpt-6-astra` | 39 | 1.05M | Medium | High | Exceptional *(reasoning)* | Yes |
| **GPT-6.1 Sol** | `gpt-6.1-sol` | 8 | 1.05M | Medium | Medium | Exceptional *(reasoning)* | Yes |
| **GPT-6 Sol** | `gpt-6-sol` | 8 | 1.05M | Medium | Medium | Exceptional *(reasoning)* | Yes |
| **GPT-6 Luna** | `gpt-6-luna` | 1 | 1.05M | Fast | Low | Excellent *(reasoning)* | Yes |
| **GPT-5.6 Sol** | `gpt-5.6-sol` | 16 | 1.05M | Medium | High | Exceptional *(reasoning)* | Yes |
| **GPT-5.6 Terra** | `gpt-5.6-terra` | 8 | 1.05M | Medium | Medium | Exceptional *(reasoning)* | Yes |
| **GPT-5.6 Luna** | `gpt-5.6-luna` | 1 | 1.05M | Fast | Low | Excellent *(reasoning)* | Yes |
| **GPT-5.5** | `gpt-5.5` | 20 | 1.05M | Medium | High | Exceptional *(reasoning)* | Yes |
| **GPT-5.4** | `gpt-5.4` | 10 | 1M | Medium | High | Exceptional *(reasoning)* | Yes |
| **GPT-5.2** | `gpt-5.2` | 8 | 400K | Medium | High | Exceptional *(reasoning)* | Yes |
| **GPT-5** | `gpt-5` | 6 | 200K | Medium | High | Exceptional *(reasoning)* | Yes |
| **GPT-5.4 Mini** | `gpt-5.4-mini` | 3 | 400K | Fast | Medium | Excellent *(reasoning)* | Yes |
| **GPT-5 Mini** | `gpt-5-mini` | 2 | 200K | Fast | Medium | Excellent *(reasoning)* | Yes |
| **GPT-5 Nano** | `gpt-5-nano` | 1 | 128K | Very\_fast | Low | Good *(reasoning)* | Yes |

### Google (Gemini) Models

| Model | Model id | Credits | Context | Speed | Cost | Reasoning | Vision |
| - | - | - | - | - | - | - | - |
| **Gemini 3.7 Flash** | `gemini-3.7-flash` | 3 | 1.05M | Fast | Medium | Exceptional *(reasoning)* | Yes |
| **Gemini 3.6 Flash** | `gemini-3.6-flash` | 3 | 1M | Fast | Medium | Excellent *(reasoning)* | Yes |
| **Gemini 3.5 Flash** | `gemini-3.5-flash` | 6 | 1M | Fast | Medium | Exceptional *(reasoning)* | Yes |
| **Gemini 3.5 Flash Lite** | `gemini-3.5-flash-lite` | 2 | 1M | Very\_fast | Very\_low | Good *(reasoning)* | Yes |
| **Gemini 3.1 Pro** | `gemini-3.1-pro-preview` | 8 | 1M | Medium | High | Exceptional *(reasoning)* | Yes |
| **Gemini 2.5 Flash** | `gemini-2.5-flash` | 2 | 1M | Fast | Medium | Excellent *(reasoning)* | Yes |
| **Gemini 2.5 Flash Lite** | `gemini-2.5-flash-lite` | 1 | 1M | Very\_fast | Very\_low | Good *(reasoning)* | Yes |

### Anthropic (Claude) Models

| Model | Model id | Credits | Context | Speed | Cost | Reasoning | Vision |
| - | - | - | - | - | - | - | - |
| **Claude Fable 5.1** | `claude-fable-5-1` | 35 | 1M | Slow | High | Exceptional *(reasoning)* | Yes |
| **Claude Opus 5.5** | `claude-opus-5-5` | 15 | 1M | Medium | High | Exceptional *(reasoning)* | Yes |
| **Claude Fable 5** | `claude-fable-5` | 39 | 1M | Slow | High | Exceptional *(reasoning)* | Yes |
| **Claude Opus 5** | `claude-opus-5` | 20 | 1M | Medium | High | Exceptional *(reasoning)* | Yes |
| **Claude Opus 4.8** | `claude-opus-4-8` | 20 | 1M | Medium | High | Exceptional *(reasoning)* | Yes |
| **Claude Sonnet 5.5** | `claude-sonnet-5-5` | 8 | 1M | Fast | Medium | Exceptional *(reasoning)* | Yes |
| **Claude Sonnet 5** | `claude-sonnet-5` | 8 | 1M | Fast | Medium | Exceptional *(reasoning)* | Yes |
| **Claude Opus 4.7** | `claude-opus-4-7` | 20 | 1M | Medium | High | Exceptional *(reasoning)* | Yes |
| **Claude Opus 4.6** | `claude-opus-4-6` | 20 | 1M | Medium | High | Exceptional *(reasoning)* | Yes |
| **Claude Sonnet 4.6** | `claude-sonnet-4-6` | 12 | 1M | Fast | Medium | Excellent *(reasoning)* | Yes |
| **Claude Haiku 4.5** | `claude-haiku-4-5` | 4 | 200K | Fast | Low | Good | Yes |
| **Claude Opus 4.5** | `claude-opus-4-5` | 20 | 200K | Medium | High | Exceptional | Yes |

### xAI (Grok) Models

| Model | Model id | Credits | Context | Speed | Cost | Reasoning | Vision |
| - | - | - | - | - | - | - | - |
| **Grok 4.7** | `grok-4.7` | 9 | 500K | Medium | Medium | Exceptional *(reasoning)* | Yes |
| **Grok 4.6** | `grok-4.6` | 9 | 500K | Medium | Medium | Exceptional *(reasoning)* | Yes |
| **Grok 4.5** | `grok-4.5` | 8 | 500K | Medium | Medium | Exceptional *(reasoning)* | Yes |
| **Grok 4.3** | `grok-4.3` | 5 | 1M | Medium | Low | Excellent *(reasoning)* | Yes |
| **Grok 4.20 Reasoning** | `grok-4.20-0309-reasoning` | 5 | 1M | Medium | Low | Excellent *(reasoning)* | Yes |
| **Grok 4.20 Non-Reasoning** | `grok-4.20-0309-non-reasoning` | 5 | 1M | Fast | Low | Good | Yes |
| **Grok Build 0.1** | `grok-build-0.1` | 4 | 256K | Fast | Low | Good | Yes |

### Zhipu AI (GLM) Models

| Model | Model id | Credits | Context | Speed | Cost | Reasoning | Vision |
| - | - | - | - | - | - | - | - |
| **GLM-5.3** | `glm-5.3` | 6 | 1M | Medium | Low | Exceptional *(reasoning)* | — |
| **GLM-5.3 Flash** | `glm-5.3-flash` | 1 | 1M | Fast | Low | Excellent *(reasoning)* | — |
| **GLM-5.2** | `glm-5.2` | 6 | 1M | Medium | Low | Exceptional *(reasoning)* | — |
| **GLM-4.6** | `glm-4.6` | 3 | 200K | Medium | Low | Excellent | — |
| **GLM-4.7** | `glm-4.7` | 3 | 200K | Medium | Low | Excellent *(reasoning)* | — |
| **GLM-4.7 Flash** | `glm-4.7-flash` | 0 | 200K | Very\_fast | Very\_low | Good | — |
| **GLM-5.1** | `glm-5.1` | 6 | 200K | Medium | Low | Exceptional *(reasoning)* | — |
| **GLM-4.6V** | `glm-4.6v` | 2 | 128K | Medium | Low | Excellent | Yes |

### MiniMax Models

| Model | Model id | Credits | Context | Speed | Cost | Reasoning | Vision |
| - | - | - | - | - | - | - | - |
| **MiniMax M2.7 Highspeed** | `MiniMax-M2.7-highspeed` | 3 | 205K | Very fast | Low | Excellent *(reasoning)* | — |
| **MiniMax M2.5 Highspeed** | `MiniMax-M2.5-highspeed` | 3 | 205K | Very fast | Low | Excellent *(reasoning)* | — |
| **MiniMax M3** | `MiniMax-M3` | 2 | 1M | Medium | Low | Exceptional *(reasoning)* | Yes |
| **MiniMax M2.7** | `MiniMax-M2.7` | 2 | 205K | Medium | Low | Excellent *(reasoning)* | — |
| **MiniMax M2.5** | `MiniMax-M2.5` | 2 | 205K | Fast | Very\_low | Excellent *(reasoning)* | — |

### DeepSeek Models

| Model | Model id | Credits | Context | Speed | Cost | Reasoning | Vision |
| - | - | - | - | - | - | - | - |
| **DeepSeek V4 Flash** | `deepseek-v4-flash` | 2 | 1M | Fast | Low | Good *(reasoning)* | — |
| **DeepSeek V4 Flash Vision** | `deepseek-v4-flash-vision-exp` | 2 | 1M | Fast | Low | Excellent *(reasoning)* | Yes |
| **DeepSeek V4 Pro** | `deepseek-v4-pro` | 5 | 1M | Medium | Low | Excellent *(reasoning)* | — |

### Moonshot AI (Kimi) Models

| Model | Model id | Credits | Context | Speed | Cost | Reasoning | Vision |
| - | - | - | - | - | - | - | - |
| **Kimi K3** | `kimi-k3` | 12 | 1M | Medium | High | Exceptional *(reasoning)* | Yes |
| **Kimi K2.7 Code** | `kimi-k2.7-code` | 5 | 256K | Medium | Medium | Exceptional *(reasoning)* | — |
| **Kimi K2.7 Code Highspeed** | `kimi-k2.7-code-highspeed` | 9 | 256K | Fast | High | Exceptional *(reasoning)* | — |
| **Kimi K2.6** | `kimi-k2.6` | 4 | 256K | Medium | Medium | Exceptional *(reasoning)* | Yes |

### Mistral AI Models

| Model | Model id | Credits | Context | Speed | Cost | Reasoning | Vision |
| - | - | - | - | - | - | - | - |
| **Mistral Small** | `mistral-small-latest` | 1 | 32K | Fast | Low | Good | Yes |
| **Mistral Medium** | `mistral-medium-latest` | 6 | 32K | Medium | Medium | Excellent | — |
| **Mistral Large** | `mistral-large-latest` | 2 | 128K | Medium | Medium | Excellent | — |

### Cohere Models

| Model | Model id | Credits | Context | Speed | Cost | Reasoning | Vision |
| - | - | - | - | - | - | - | - |
| **Command A** | `command-a-03-2025` | 21 | 256K | Medium | High | Exceptional | — |
| **Command R+** | `command-r-plus-08-2024` | 21 | 128K | Medium | High | Exceptional | — |
| **Command R** | `command-r-08-2024` | 2 | 128K | Fast | Low | Excellent | — |
| **Command R7B** | `command-r7b-12-2024` | 1 | 128K | Very\_fast | Low | Excellent | — |
| **Command A Translate** | `command-a-translate-08-2025` | 21 | 8K | Fast | Low | Good | — |

### Groq Models

| Model | Model id | Credits | Context | Speed | Cost | Reasoning | Vision |
| - | - | - | - | - | - | - | - |
| **GPT-OSS (20B)** | `openai/gpt-oss-20b` | 1 | 131K | Extremely\_fast | Very\_low | Good | — |
| **Qwen 3.8 (27B)** | `qwen/qwen3.8-27b` | 7 | 131K | Very fast | Low | Excellent | — |
| **Groq Compound** | `groq/compound` | 2 | 131K | Very fast | Low | Excellent | — |
| **GPT-OSS (120B)** | `openai/gpt-oss-120b` | 2 | 131K | Fast | Low | Excellent | — |

### Fireworks AI Models

| Model | Model id | Credits | Context | Speed | Cost | Reasoning | Vision |
| - | - | - | - | - | - | - | - |
| **DeepSeek V3** | `deepseek-v3` | 4 | 64K | Fast | Low | Good | — |
| **DeepSeek R1** | `deepseek-r1` | 25 | 64K | Medium | Low | Excellent *(reasoning)* | — |
| **Llama V3 (8B)** | `llama-v3-8b-instruct` | 2 | 8K | Very\_fast | Very\_low | Good | — |
| **Llama V3 (70B)** | `llama-v3-70b-instruct` | 8 | 8K | Medium | Medium | Excellent | — |
| **Llama V3.1 (8B)** | `llama-v3p1-8b-instruct` | 2 | 128K | Very\_fast | Very\_low | Good | — |
| **Llama V3.1 (70B)** | `llama-v3p1-70b-instruct` | 8 | 128K | Fast | Low | Excellent | — |
| **Llama V3.1 (405B)** | `llama-v3p1-405b-instruct` | 8 | 128K | Slow | High | Exceptional | — |
| **Yi Large** | `yi-large` | 8 | 32K | Medium | Medium | Excellent | — |
| **Mixtral (8x7B)** | `mixtral-8x7b-instruct` | 4 | 32K | Fast | Low | Good | — |
| **Mixtral (8x22B)** | `mixtral-8x22b-instruct` | 10 | 64K | Medium | Medium | Excellent | — |
| **Qwen 2.5 (72B)** | `qwen2p5-72b-instruct` | 8 | 128K | Medium | Medium | Excellent | — |

### Together AI Models

| Model | Model id | Credits | Context | Speed | Cost | Reasoning | Vision |
| - | - | - | - | - | - | - | - |
| **GLM-5.3** | `zai-org/GLM-5.3` | 6 | 1M | Medium | Low | Exceptional *(reasoning)* | — |
| **GLM-5.2** | `zai-org/GLM-5.2` | 6 | 1M | Medium | Low | Exceptional *(reasoning)* | — |
| **Kimi K3** | `moonshotai/Kimi-K3` | 12 | 1M | Medium | Medium | Exceptional *(reasoning)* | — |
| **MiniMax M3** | `MiniMaxAI/MiniMax-M3` | 2 | 524K | Medium | Low | Excellent *(reasoning)* | — |
| **GPT-OSS (120B)** | `openai/gpt-oss-120b` | 2 | 131K | Fast | Low | Excellent | — |
| **Qwen 2.5 Coder (32B)** | `Qwen/Qwen2.5-Coder-32B-Instruct` | 7 | 32K | Fast | Low | Excellent | — |
| **Qwen 2.5 Turbo (72B)** | `Qwen/Qwen2.5-72B-Instruct-Turbo` | 10 | 128K | Fast | Medium | Excellent | — |

### Ollama Models (Local)

| Model | Model id | Credits | Context | Speed | Cost | Reasoning | Vision |
| - | - | - | - | - | - | - | - |
| **Llama 3.3 (70B) Local** | `llama3.3:70b` | 0 | 128K | Medium | Very\_low | Excellent | — |
| **Llama 3.1 (8B) Local** | `llama3.1:8b` | 0 | 128K | Fast | Very\_low | Good | — |
| **Llama 3.1 (70B) Local** | `llama3.1:70b` | 0 | 128K | Medium | Very\_low | Excellent | — |
| **Llama 3.1 (405B) Local** | `llama3.1:405b` | 0 | 128K | Slow | Very\_low | Exceptional | — |

## Comparison by Category

### Best for Speed

> [accordion] Very fast
>
> * **GPT-5 Nano** - 1 credit
> * **Gemini 3.5 Flash Lite** - 2 credits
> * **GPT-OSS (20B)** - 1 credit
> * **Mistral Small** - 1 credit

> [accordion] Fast
>
> * **GPT-6 Luna** - 1 credit
> * **Gemini 3.8 Flash** - 3 credits
> * **Claude Haiku 4.5** - 4 credits
> * **Claude Sonnet 5.5** - 8 credits

> [accordion] Medium (deep reasoning)
>
> * **GPT-6.1 Sol** - 8 credits
> * **Claude Opus 5.5** - 15 credits
> * **GPT-6 Astra** - 39 credits
> * **Claude Fable 5.1** - 35 credits

### Best for Cost

> [accordion] 1 credit
>
> * **GPT-6 Luna** - 1 credit
> * **GPT-5 Nano** - 1 credit
> * **Gemini 3.5 Flash Lite** - 2 credits
> * **GLM-5.3 Flash** - 1 credit
> * **Mistral Small** - 1 credit

> [accordion] 2-5 credits
>
> * **DeepSeek V4 Flash** - 2 credits
> * **MiniMax M3** - 2 credits
> * **Gemini 3.8 Flash** - 3 credits
> * **Claude Haiku 4.5** - 4 credits
> * **DeepSeek V4 Pro** - 5 credits

> [accordion] Premium
>
> * **GPT-6.1 Sol** - 8 credits
> * **Claude Sonnet 5.5** - 8 credits
> * **Claude Opus 5.5** - 15 credits
> * **Claude Fable 5.1** - 35 credits
> * **GPT-6 Astra** - 39 credits

### Best for Context Window

> [accordion] 1M+ tokens
>
> * **GPT-6 Luna** - 1 credit (1.05M)
> * **GPT-6.1 Sol** - 8 credits (1.05M)
> * **Gemini 3.8 Flash** - 3 credits (1M)
> * **Claude Sonnet 5.5** - 8 credits (1M)
> * **DeepSeek V4 Flash** - 2 credits (1M)

> [accordion] 200K-500K tokens
>
> * **Grok 4.7** - 9 credits (500K)
> * **GPT-5.4 Mini** - 3 credits (400K)
> * **Claude Haiku 4.5** - 4 credits (200K)

### Best for Reasoning

> [accordion] Exceptional Reasoning
>
> * **GPT-6 Astra** - 39 credits
> * **Claude Fable 5.1** - 35 credits
> * **Claude Opus 5.5** - 15 credits
> * **GPT-6.1 Sol** - 8 credits
> * **Claude Sonnet 5.5** - 8 credits

> [accordion] Excellent Reasoning
>
> * **GPT-6 Luna** - 1 credit
> * **Gemini 3.8 Flash** - 3 credits
> * **DeepSeek V4 Pro** - 5 credits
> * **GPT-5.4 Mini** - 3 credits

> [accordion] Good Reasoning
>
> * **GPT-5 Nano** - 1 credit
> * **Claude Haiku 4.5** - 4 credits
> * **Mistral Small** - 1 credit

## Use Case Recommendations

### Customer Support Chatbots

**Recommended:**

* **GPT-6 Luna** (1 credit) - Best balance
* **Gemini 3.8 Flash** (3 credits) - Stronger reasoning
* **Claude Haiku 4.5** (4 credits) - Fast and careful

**Budget Option:**

* **GPT-5 Nano** (1 credit)
* **Gemini 3.5 Flash Lite** (2 credits)

### Code Analysis & Debugging

**Recommended:**

* **GPT-6.1 Sol** (8 credits) - Deep reasoning
* **Claude Sonnet 5.5** (8 credits) - Agentic coding
* **DeepSeek V4 Pro** (5 credits) - Cost-effective

**Premium:**

* **Claude Opus 5.5** (15 credits) - Long-running work

### Content Generation

**Recommended:**

* **Claude Sonnet 5.5** (8 credits) - Nuanced writing
* **GPT-6 Luna** (1 credit) - Balanced

**Budget Option:**

* **Gemini 3.5 Flash Lite** (2 credits)

### Long Document Analysis

**Recommended:**

* **GPT-6 Luna** (1 credit) - 1.05M context
* **Gemini 3.8 Flash** (3 credits) - 1M context

**Alternative:**

* **Claude Sonnet 5.5** (8 credits) - 1M context

### Real-Time Applications

**Recommended:**

* **GPT-OSS (20B)** (1 credit) - Fast on Groq
* **GPT-5 Nano** (1 credit)
* **Gemini 3.5 Flash Lite** (2 credits)

### High-Volume Production

**Recommended:**

* **GPT-6 Luna** (1 credit)
* **GLM-5.3 Flash** (1 credit)
* **Mistral Small** (1 credit)

### Privacy-Sensitive Applications

**Recommended:**

* **Ollama models** (local, no API calls)
* Use BYOK with any provider

## Provider Comparison

| Provider | Models | Credits per step | Strength |
| - | - | - | - |
| **OpenAI** | 14 | 1-39 credits | Most comprehensive |
| **Google** | 8 | 1-8 credits | Long context, fast |
| **Anthropic** | 12 | 4-39 credits | Agentic coding and writing |
| **xAI** | 7 | 4-9 credits | Real-time knowledge |
| **DeepSeek** | 3 | 2-5 credits | Best value reasoning |
| **Mistral** | 3 | 1-6 credits | Open and efficient |
| **Cohere** | 5 | 1-21 credits | Enterprise-grade |
| **Groq** | 4 | 1-7 credits | Speed champion |

## Decision Tree

```
Need deep reasoning?
├─ Yes → Budget?
│  ├─ Premium → Claude Opus 5.5 (15 credits)
│  └─ Budget → DeepSeek V4 Pro (5 credits)
└─ No → Need speed above all?
   ├─ Yes → GPT-5 Nano (1 credit)
   └─ Balanced → GPT-6 Luna (1 credit) ⭐ Best Overall
```

## Next Steps

> [card] [Provider Overview](/docs/providers/overview.md)
>
> Detailed provider information

> [card] [Reasoning Models](/docs/providers/reasoning-models.md)
>
> Deep dive into reasoning

> [card] [Bring Your Own Keys](/docs/providers/bring-your-own-keys.md)
>
> Use your own API keys

> [card] [Get Started](/docs/developers/quickstart.md)
>
> Build your first bot


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.