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

# Models & Pricing

> 20+ models behind one endpoint — open-weight and frontier — with transparent, live-synced pricing.

## Live catalog

The source of truth is the public catalog endpoint — no auth required:

```bash theme={null}
curl https://gw.hyreagent.fun/api/inference/v1/models
```

Each entry includes the model id, tier, context length, live per-1M-token pricing (human-readable and micro-USD integers), whether it's today's free-tier model, and any active promo:

```json theme={null}
{
  "id": "zai-org/GLM-5",
  "name": "GLM 5",
  "tier": "STANDARD",
  "provider": "deepinfra",
  "context_length": 202752,
  "free_eligible": false,
  "pricing": {
    "input_per_1m": "$0.72",
    "output_per_1m": "$2.496",
    "input_per_1m_micro": 720000,
    "output_per_1m_micro": 2496000
  }
}
```

<Note>
  Prices are synced from upstream daily and can carry limited-time promo discounts, so always read live pricing from `GET /models` rather than hardcoding it. Docs tables below list the lineup, not prices.
</Note>

## Model lineup

### Open-weight models

Served via DeepInfra. Tiers by retail output price: **FLASH** (under \$0.60/M), **STANDARD** (under \$3/M), **PREMIUM** (\$3/M and above).

| Model id                                            | Name                  | Tier     | Context |
| --------------------------------------------------- | --------------------- | -------- | ------- |
| `deepseek-ai/DeepSeek-V4-Flash`                     | DeepSeek V4 Flash     | FLASH    | 1M      |
| `zai-org/GLM-4.7-Flash`                             | GLM 4.7 Flash         | FLASH    | 200K    |
| `stepfun-ai/Step-3.5-Flash`                         | Step 3.5 Flash        | FLASH    | 262K    |
| `google/gemma-4-31B-it`                             | Gemma 4 31B           | FLASH    | 262K    |
| `zai-org/GLM-5`                                     | GLM 5                 | STANDARD | 200K    |
| `Qwen/Qwen3.5-397B-A17B`                            | Qwen 3.5 397B         | STANDARD | 262K    |
| `meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8` | Llama 4 Maverick      | STANDARD | 1M      |
| `MiniMaxAI/MiniMax-M2.7`                            | MiniMax M2.7          | STANDARD | 196K    |
| `nvidia/NVIDIA-Nemotron-3-Super-120B-A12B`          | Nemotron 3 Super 120B | STANDARD | 262K    |
| `deepseek-ai/DeepSeek-V4-Pro`                       | DeepSeek V4 Pro       | PREMIUM  | 1M      |
| `moonshotai/Kimi-K2.6`                              | Kimi K2.6             | PREMIUM  | 262K    |
| `moonshotai/Kimi-K2.7-Code`                         | Kimi K2.7             | PREMIUM  | 262K    |
| `zai-org/GLM-5.2`                                   | GLM 5.2               | PREMIUM  | 1M      |

### Frontier models

Proprietary frontier models served through HYRE's compute partners — same endpoint, same key. All PREMIUM tier.

| Model id                        | Name              | Context |
| ------------------------------- | ----------------- | ------- |
| `anthropic-claude-opus-4-8`     | Claude Opus 4.8   | 1M      |
| `anthropic-claude-sonnet-4-6`   | Claude Sonnet 4.6 | 1M      |
| `openai-gpt-55`                 | GPT-5.5           | 1M      |
| `openai-gpt-54`                 | GPT-5.4           | 1M      |
| `openai-gpt-54-mini`            | GPT-5.4 Mini      | 400K    |
| `google-gemini-3-1-pro-preview` | Gemini 3.1 Pro    | 1M      |
| `google-gemini-3-flash-preview` | Gemini 3 Flash    | 256K    |
| `x-ai-grok-4-3`                 | Grok 4.3          | 1M      |

## How pricing works

* **Retail = live upstream price × 1.20.** No hidden margins, no per-seat fees — you pay a flat 20% markup over the provider rate for routing, billing, and support.
* Billing is **per token** (input and output priced separately), computed from the usage the upstream reports. Streaming responses bill from the final usage frame.
* Metered keys **reserve against your balance before the call and settle to actual usage after** — you're never charged more than the tokens you consumed.
* **Promos** (limited-time discounts, holder free tiers) apply automatically and are visible in `GET /models` while active — no code changes needed on your side.

## Free daily model

One FLASH/STANDARD model is hand-picked as free each UTC day (`free_eligible: true` in `/models`), with a per-user quota for eligible \$HYRE holders. Treat it as a bonus, not capacity to build against.