Mercury 2.5
Mercury 2.5 is the fastest reasoning LLM, and the latest diffusion LLM (dLLM) from Inception. Instead of generating tokens sequentially, Mercury 2.5 produces and refines multiple tokens in parallel, achieving...
Input / 1M
$0.04
Output / 1M
$0.15
Context
260K
AA Intelligence
12.3
Pricing
via OpenRouter · USD per 1M tokens · synced 2026-10-24$0.04
input / 1M
$0.15
output / 1M
$0.004
cached input / 1M
$0.19
1M in + 1M out
- Context window
- 260,000 tokens
- Max output
- 65,536 tokens
- Providers
- 1
- Cheapest provider
- Inception · $0.04 / $0.15
- Best throughput
- —
- AA Intelligence / Coding / Agentic
- 12.3 / — / —
✓ Tool calling
✓ Reasoning
✓ Structured outputs
✗ Image input
✗ File input
✗ Audio input
✓ Prompt caching
Price history
tracking started — chart appears after 7 days of snapshots (6 recorded)Current: $0.04 input · $0.15 output per 1M tokens
Benchmarks
Third-party indices as published through OpenRouter — Artificial Analysis and Design Arena. Not measured by us.
- AA Intelligence Index
- 12.3
- AA Coding Index
- —
- AA Agentic Index
- —
Providers
1 provider · per 1M tokens · last 30 min metrics| Provider | Quant | Context | Input | Output | Uptime | Tok/s |
|---|---|---|---|---|---|---|
| Inception | unknown | 260K | $0.04 | $0.15 | 100% | — |
Provider data from OpenRouter's endpoint listing, refreshed daily.
Specifications
- Organisation
- inception
- Released on OpenRouter
- 2026-09-08
- Context window
- 260,000 tokens
- Max output
- 65,536 tokens
- Input
- text
- Output
- text
- Tokenizer
- Other
- Moderated by provider
- No
- OpenRouter id
- inception/mercury-2.5
Compare with
Popular comparisons
Qwen3-235B-A22B-Instruct-2507 open
$0.21
Kimi-K2-Instruct open
$1.00
Qwen3-235B-A22B-Thinking-2507 open
$1.21
Qwen3-Coder-480B-A35B-Instruct open
$0.48
GLM-4.5 open
$1.00
NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16 open
$0.10
GPT-5 Nano
$0.14
Gemini 2.5 Flash Lite
$0.18
GPT-4.1 Nano
$0.18
Nova Lite 1.0
$0.11
Prices, context and providers from the public OpenRouter API, snapshotted daily; benchmark indices from Artificial Analysis. See /methodology.