Mercury 2.5 Preview
inception/mercury-2.5-previewMercury 2.5 Preview is Inception's latest and most intelligent diffusion language model. Instead of generating tokens strictly one at a time, it produces and refines multiple tokens in parallel, reaching up to 1,107 tokens per second on standard GPUs. It delivers a 10+ point intelligence gain over Mercury 2, with tunable reasoning, parallel tool calls, schema-aligned JSON output, and a 260K context window. It is built for latency-sensitive production work such as search agents, voice pipelines, customer support, rapid coding iteration, and coding subagents.
Context
260K
260,000 tokens
Max output
66K
65,536 tokens
Input
$0.04
per 1M tokens
Output
$0.15
per 1M tokens
Providers
1
1 listings
Intelligence
—
Not yet scored
Pricing
USD per 1M tokens · listed by NanoGPT
| Token type | Standard |
|---|---|
| Input | $0.04 |
| Output | $0.15 |
| Cache read | $0.004 |
Lowest input price across providers: $0.04.
Capabilities
Effort: none, low, medium, high
Input
Output
Benchmarks
Published scores and Design Arena Elo
No published benchmark scores for this model yet.
Rankings
Position on each OneAtlas leaderboard
Providers · 1
1 listings
| Provider | Model ID | Context | Input limit | Output | Input $ | Output $ | Reasoning $ | Cache read | Cache write | Audio in | Audio out | Reasoning | Tools | Structured | Temp. | Updated |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
 NanoGPT | inception/mercury-2.5-preview | 260K | 260K | 66K | $0.04 | $0.15 | — | $0.004 | — | — | — | Yes | Yes | Yes | Unknown | Sep 1, 2026 |