> ## Documentation Index
> Fetch the complete documentation index at: https://docs.prysm1.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Model catalog

> The 27 models PRYSM routes across, with live pricing, context windows, and strengths — grouped by tier.

PRYSM routes across **27 models from 12 providers** (OpenAI, Anthropic, Google, DeepSeek,
xAI, Mistral, Moonshot, Alibaba, Perplexity, Meta, Zhipu, SambaNova). With `model: "auto"`, PRYSM picks the best-value model
for each prompt; pass any `id` below to pin a single request to that model.

<Note>
  Prices are **USD per million tokens (MTok)**. This page mirrors the live catalog —
  fetch the current data any time from [`GET /v1/models`](/api-reference/models) or
  `prysm models`.
</Note>

## Budget

Fast and inexpensive — ideal for classification, simple Q\&A, and high-volume work.

| ID                      | Model                 | Provider | Context | Input  | Output | Strengths                                           |
| ----------------------- | --------------------- | -------- | ------- | ------ | ------ | --------------------------------------------------- |
| `mistral-nemo`          | Mistral Nemo          | mistral  | 128K    | \$0.02 | \$0.02 | classification, simple, fast                        |
| `gpt-5-nano`            | GPT-5 Nano            | openai   | 128K    | \$0.05 | \$0.40 | simple, chat, fast                                  |
| `gemini-2.5-flash-lite` | Gemini 2.5 Flash-Lite | google   | 1M      | \$0.10 | \$0.40 | simple, translate, multimodal                       |
| `deepseek-v4-flash`     | DeepSeek V4 Flash     | deepseek | 1M      | \$0.14 | \$0.28 | general, code, math, reasoning, multilingual, value |
| `llama-4-maverick`      | Llama 4 Maverick      | meta     | 128K    | \$0.27 | \$0.85 | general, multilingual, open-source                  |

## Mid

The everyday workhorses — strong quality at a sensible price across most tasks.

| ID                 | Model            | Provider   | Context | Input  | Output | Strengths                          |
| ------------------ | ---------------- | ---------- | ------- | ------ | ------ | ---------------------------------- |
| `gpt-oss-20b`      | GPT-OSS 20B      | openrouter | 128K    | \$0.10 | \$0.40 | math, reasoning, code, open-source |
| `grok-4.1`         | Grok 4.1         | xai        | 2M      | \$0.20 | \$0.50 | realtime, news, long-context       |
| `glm-4.5-air`      | GLM-4.5 Air      | zhipu      | 128K    | \$0.20 | \$1.10 | code, general, value               |
| `gpt-5-mini`       | GPT-5 Mini       | openai     | 128K    | \$0.25 | \$2.00 | structured, json, tool-use         |
| `qwen-2.5-72b`     | Qwen 2.5 72B     | alibaba    | 128K    | \$0.30 | \$0.80 | math, chinese, multilingual, code  |
| `kimi-k2`          | Kimi K2          | moonshot   | 128K    | \$0.35 | \$1.50 | long-context, chinese, analysis    |
| `mistral-medium-3` | Mistral Medium 3 | mistral    | 131K    | \$0.40 | \$2.00 | multilingual, european, code       |
| `gemini-3-flash`   | Gemini 3 Flash   | google     | 1M      | \$0.50 | \$3.00 | multimodal, fast, long-context     |
| `claude-haiku-4.5` | Claude Haiku 4.5 | anthropic  | 200K    | \$1.00 | \$5.00 | instruction, safety, chat, fast    |
| `sonar`            | Sonar            | perplexity | 128K    | \$1.00 | \$1.00 | realtime, news, search, citations  |

## Premium

High-accuracy reasoning, writing, and code for higher-stakes work.

| ID                  | Model             | Provider   | Context | Input  | Output  | Strengths                                       |
| ------------------- | ----------------- | ---------- | ------- | ------ | ------- | ----------------------------------------------- |
| `deepseek-v4-pro`   | DeepSeek V4 Pro   | deepseek   | 1M      | \$0.44 | \$0.87  | reasoning, math, code, chain-of-thought         |
| `kimi-k2-thinking`  | Kimi K2 Thinking  | openrouter | 256K    | \$0.60 | \$2.50  | reasoning, math, chain-of-thought, long-context |
| `glm-4.6`           | GLM-4.6           | zhipu      | 200K    | \$0.60 | \$2.20  | code, agentic, reasoning, tool-use              |
| `gpt-5.2`           | GPT-5.2           | openai     | 128K    | \$1.75 | \$14.00 | reasoning, agentic, structured, code            |
| `gemini-3.1-pro`    | Gemini 3.1 Pro    | google     | 1M      | \$2.00 | \$12.00 | multimodal, long-context, reasoning             |
| `claude-sonnet-4.5` | Claude Sonnet 4.5 | anthropic  | 1M      | \$3.00 | \$15.00 | writing, code, nuance, instruction              |
| `sonar-pro`         | Sonar Pro         | perplexity | 200K    | \$3.00 | \$15.00 | realtime, news, search, citations, deep-search  |

## Frontier

Maximum capability for the hardest problems — reserved for when accuracy is paramount.

| ID                | Model           | Provider  | Context | Input   | Output   | Strengths                          |
| ----------------- | --------------- | --------- | ------- | ------- | -------- | ---------------------------------- |
| `grok-4.1-heavy`  | Grok 4.1 Heavy  | xai       | 2M      | \$3.00  | \$15.00  | deep-reasoning, realtime, science  |
| `claude-opus-4.6` | Claude Opus 4.6 | anthropic | 1M      | \$5.00  | \$25.00  | complex-reasoning, writing, safety |
| `claude-opus-4.8` | Claude Opus 4.8 | anthropic | 1M      | \$6.00  | \$30.00  | complex-reasoning, agentic, safety |
| `gpt-5.2-pro`     | GPT-5.2 Pro     | openai    | 128K    | \$21.00 | \$168.00 | advanced-reasoning, science, legal |

## Using a specific model

Pass an `id` from the tables above instead of `auto` to skip routing:

<CodeGroup>
  ```python Python theme={null}
  client.chat.completions.create(
      model="claude-sonnet-4.5",
      messages=[{"role": "user", "content": "Draft a launch announcement"}],
  )
  ```

  ```typescript Node theme={null}
  await client.chat.completions.create({
    model: "claude-sonnet-4.5",
    messages: [{ role: "user", content: "Draft a launch announcement" }],
  });
  ```

  ```bash cURL theme={null}
  curl https://api.prysm1.com/v1/chat/completions \
    -H "Authorization: Bearer $PRYSM_API_KEY" \
    -H "Content-Type: application/json" \
    -d '{ "model": "claude-sonnet-4.5", "messages": [{ "role": "user", "content": "Draft a launch announcement" }] }'
  ```
</CodeGroup>

<Tip>
  Reference these IDs in a [BRAIN.md](/concepts/brain-md) `model`, `rules`, `blocked`, or
  `fallback` field to shape routing without touching code.
</Tip>