Mercury 2.5 is the most capable diffusion LLM, delivering high-speed, low-cost reasoning with 260K context and tool calling — ideal for agents, search, and coding.
curl https://api.venice.ai/api/v1/chat/completions \
-H "Authorization: Bearer $VENICE_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "mercury-2-5",
"messages": [{ "role": "user", "content": "Build without permission." }]
}'Overview
Mercury 2.5 is Inception's most capable production diffusion LLM, released in September 2026. It generates and refines tokens in parallel, achieving speeds over 1,100 tokens per second. With a 260K context window, function calling, and web search, it excels in latency-sensitive workflows like agents, voice, and coding.
On Venice, Mercury 2.5 runs with full prompt anonymity — your inputs are never stored or profiled. This enables private, uncensored use of a high-performance reasoning model in production. The combination of zero retention and access to web search and tool use makes it ideal for developers building permissionless agents without surveillance trade-offs.
Agent quickstart
The API is OpenAI-compatible: change the base URL and the model id and existing client code works unchanged.
curl https://api.venice.ai/api/v1/chat/completions \
-H "Authorization: Bearer $VENICE_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "mercury-2-5",
"stream": true,
"messages": [{ "role": "user", "content": "Draft the release note." }]
}'curl https://api.venice.ai/api/v1/chat/completions \
-H "Authorization: Bearer $VENICE_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "mercury-2-5",
"messages": [{ "role": "user", "content": "Find the rate limits." }],
"tools": [{
"type": "function",
"function": {
"name": "search_docs",
"description": "Search the API documentation.",
"parameters": {
"type": "object",
"properties": { "query": { "type": "string" } },
"required": ["query"]
}
}
}],
"tool_choice": "auto"
}'import os
from openai import OpenAI
client = OpenAI(
api_key=os.environ["VENICE_API_KEY"],
base_url="https://api.venice.ai/api/v1",
)
resp = client.chat.completions.create(
model="mercury-2-5",
messages=[{"role": "user", "content": "Build without permission."}],
)
print(resp.choices[0].message.content)Specifications
Assessment
Use cases
Prompting
Use explicit JSON schema in your prompt to ensure structured output.
Break complex tasks into parallel tool calls to leverage Mercury 2.5’s architecture.
Leverage web search by including 'search:' in your query for real-time data.
Keep prompts under 260K tokens to maximize context for reasoning.
Use clear reasoning directives like 'think step by step' to improve output quality.
Test with the promotional pricing tier to evaluate cost-performance before scaling.
Alternatives
| Model | Best for | Context | Open weights | Price (Venice) |
|---|---|---|---|---|
| Mercury 2.5 | Fast reasoning agents | 260K tokens | No | $0.25 in · $0.94 out / 1M |
| Claude Opus 5 | Deep reasoning | 1M tokens | No | $6 in · $30 out / 1M |
| DeepSeek V4 Flash 0731 | Cost-efficient throughput | 1M tokens | Yes | $0.17 in · $0.35 out / 1M |
| Google Gemma 4 31B Instruct | Open-weight reasoning | 256K tokens | Yes | $0.12 in · $0.36 out / 1M |
Mercury 2.5 is the right pick when you need the fastest reasoning with tool use and web search at low cost — outperforming rivals in agent speed and privacy without sacrificing intelligence.
Pricing
Billed per token on Venice: $0.25 per 1M input tokens and $0.94 per 1M output tokens.
New Venice accounts include a free daily allowance and 500 welcome credits — no credit card required.
Getting a key
Create a key in API settings. Nothing else is required to start.
Export it as VENICE_API_KEY so the snippets above run unedited.
Point an existing OpenAI client at https://api.venice.ai/api/v1. The scheme is part of the value: an OpenAI client given a bare host does not resolve it.
Pass mercury-2-5 as the model and send the request.
FAQ
Mercury 2.5 is Inception's most capable production diffusion LLM, released in September 2026. It generates tokens in parallel, achieving speeds over 1,100 tokens per second, with a 260K context window, tool calling, and web search — ideal for agents, coding, and search.
On Venice, Mercury 2.5 costs $0.25 per million input tokens and $0.94 per million output tokens. Cached input is $0.03 per million. A promotional rate of $0.04 input and $0.15 output is available at launch.
No. Mercury 2.5 is a proprietary model by Inception and is not open source. It is not free, but Venice offers a promotional rate at launch for reduced cost.
Yes. Mercury 2.5 supports tool use and function calling, enabling it to interact with external systems, APIs, and databases as part of agentic workflows.
Mercury 2.5 supports a 260K-token context window, allowing it to process and reason over very long documents and complex inputs.
Yes. Mercury 2.5 supports web search, allowing it to retrieve and incorporate real-time information into its responses.
Mercury 2.5 is faster and cheaper, with parallel token generation and web search, making it better for latency-sensitive agents. Claude Opus 5 offers deeper reasoning and 1M context but at a much higher cost.
Yes. On Venice, Mercury 2.5 runs under an anonymized privacy tier — your prompts are not stored, profiled, or used for training, ensuring full sovereignty over your inputs.
One key, free to start, no credit card.